SQL 2005 Customer Evidence Contact: Ram Ramanathan.

26
SQL 2005 Customer SQL 2005 Customer Evidence Evidence Contact: Ram Ramanathan Contact: Ram Ramanathan
  • date post

    19-Dec-2015
  • Category

    Documents

  • view

    221
  • download

    2

Transcript of SQL 2005 Customer Evidence Contact: Ram Ramanathan.

Page 1: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

SQL 2005 Customer SQL 2005 Customer EvidenceEvidenceSQL 2005 Customer SQL 2005 Customer EvidenceEvidence

Contact: Ram Ramanathan Contact: Ram Ramanathan

Page 2: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

SQL Server 2005 TB+ SQL Server 2005 TB+ CustomersCustomersData Warehousing w/ Relational QueryData Warehousing w/ Relational Query

5.3 TB Credit Card DW, 5 Mil card holders5.3 TB Credit Card DW, 5 Mil card holders300 users, OLAP, SSIS, SSRS, Office BSM300 users, OLAP, SSIS, SSRS, Office BSM

4 TB Consumer Packaged Goods (CPG) DW4 TB Consumer Packaged Goods (CPG) DW350 users, complex relational query, SSIS & SSRS350 users, complex relational query, SSIS & SSRS

5 TB Retail Data Warehouse on HP Superdome5 TB Retail Data Warehouse on HP SuperdomeUses SQL Server OLAP, SSIS, SSRSUses SQL Server OLAP, SSIS, SSRS

2 TB DW, originally migrated from Informix2 TB DW, originally migrated from Informix300 users, complex query, OLAP & Data Mining300 users, complex query, OLAP & Data Mining

2 TB of Clinical DW, 50 TB storage, 90 OLAP Marts2 TB of Clinical DW, 50 TB storage, 90 OLAP MartsUS Dept of Veterans Affairs, 1200 facilities US Dept of Veterans Affairs, 1200 facilities

2 TB Healthcare DW, HMC is part of Wellpoint Inc.2 TB Healthcare DW, HMC is part of Wellpoint Inc.Complex ETL, 800 SSIS packages, OLAP, RSComplex ETL, 800 SSIS packages, OLAP, RS

Page 3: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

SQL Server 2005 TB+ SQL Server 2005 TB+ CustomersCustomersGeneral VLDB & DW/BI WorkloadsGeneral VLDB & DW/BI Workloads

3 TB document management system, Sybase 3 TB document management system, Sybase movemoveState of Alaska Dept of Revenue, Permanent FundState of Alaska Dept of Revenue, Permanent Fund1 TB custom document management system 1 TB custom document management system (VS.Net)(VS.Net)State of Washington Archive Department, 600 State of Washington Archive Department, 600 usersusers

1 TB Real-time store reporting system (SSRS)1 TB Real-time store reporting system (SSRS)Data loaded continually (MSMQ) and daily (SSIS)Data loaded continually (MSMQ) and daily (SSIS)

2.8 TB SAP Business Warehouse system2.8 TB SAP Business Warehouse systemIn the process of migrating SAP R/3 OLTP to In the process of migrating SAP R/3 OLTP to SS2005SS2005

1.5 TB BASEL II Compliance DW, complex SSIS1.5 TB BASEL II Compliance DW, complex SSISNationwide Building Society, UKNationwide Building Society, UK

1 TB Staging Data Warehouse at Telecom 1 TB Staging Data Warehouse at Telecom DenmarkDenmarkCUBUS is massive BI system with up to 10,000 CUBUS is massive BI system with up to 10,000 usersusers

Page 4: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

SQL Server 2005 TB+ SQL Server 2005 TB+ CustomersCustomersData Warehousing w/ Relational QueryData Warehousing w/ Relational Query

1 TB data warehouse for CRM support1 TB data warehouse for CRM supportRuns core systems SQL Server, BASELL II on Runs core systems SQL Server, BASELL II on SS2005SS2005

1 TB gaming data warehouse, 24 casino1 TB gaming data warehouse, 24 casinoHeavy daily loads, relational queryHeavy daily loads, relational query

6 TB retail data warehouse from 700 stores, 6 TB retail data warehouse from 700 stores, OLAP, RSOLAP, RS15 TB total across multiple instances15 TB total across multiple instances

2 TB normalized DW, 250 concurrent BI users2 TB normalized DW, 250 concurrent BI usersNightly transaction loads: 900 TPS sustainedNightly transaction loads: 900 TPS sustained

5 TB retail DW, 10 TB storage, data from 1000 5 TB retail DW, 10 TB storage, data from 1000 storesstoresLargest retailer in African continent, OLAP, RS, Largest retailer in African continent, OLAP, RS, DTS DTS

3 TB relational DW, DB2 migration3 TB relational DW, DB2 migrationHeavy Complex Relational QueryHeavy Complex Relational Query

Page 5: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

SQL Server 2000 TB+ SQL Server 2000 TB+ CustomersCustomersGeneral VLDB & DW/BI WorkloadsGeneral VLDB & DW/BI Workloads20 TB European digital payment archive system20 TB European digital payment archive system

Unisys Payments Services ArchiveUnisys Payments Services Archive

13 TB Geospatial DW, to reach 25 TB13 TB Geospatial DW, to reach 25 TB4 TB data mart, 2500 field offices, 55 TB storage4 TB data mart, 2500 field offices, 55 TB storage

Multiple 1 TB instances, credit scoring analysisMultiple 1 TB instances, credit scoring analysisUses complex OLAP & SQL Server ETLUses complex OLAP & SQL Server ETL

Used by 13,000 agents in 220 officesUsed by 13,000 agents in 220 officesBuilt portal with Visual Studio .NET, OLAPBuilt portal with Visual Studio .NET, OLAP

700 restaurants, 30,000 users, 700+ concurrent700 restaurants, 30,000 users, 700+ concurrentOLTP/LOB reporting on inventory & sales dataOLTP/LOB reporting on inventory & sales data

Reporting on 1.6 TB of data, 35K+ Tx/MinReporting on 1.6 TB of data, 35K+ Tx/Min325 concurrent connections325 concurrent connections

Page 6: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

SQL Server 2000 TB+ SQL Server 2000 TB+ CustomersCustomersGeneral VLDB & DW/BI WorkloadsGeneral VLDB & DW/BI Workloads20 TB European digital payment archive system20 TB European digital payment archive system

Unisys Payments Services ArchiveUnisys Payments Services Archive

13 TB Geospatial DW, to reach 25 TB13 TB Geospatial DW, to reach 25 TB4 TB data mart, 2500 field offices, 55 TB storage4 TB data mart, 2500 field offices, 55 TB storage

1.5 TB DW, plus multiple OLAP data marts1.5 TB DW, plus multiple OLAP data martsData feeds from JD Edwards ERP systemData feeds from JD Edwards ERP system

1 TB DW, Informix migration1 TB DW, Informix migrationHeavy concurrent usage on OLAPHeavy concurrent usage on OLAP

1.7 TB DW, 2900 DW users ramping up to 50001.7 TB DW, 2900 DW users ramping up to 5000Runs on HP SuperdomeRuns on HP Superdome

6 TB Call Detail Record (CDR) DW, OLAP usage6 TB Call Detail Record (CDR) DW, OLAP usage20-way HP Superdome20-way HP Superdome

Page 7: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

SQL Server 2000 TB+ SQL Server 2000 TB+ CustomersCustomersPrivate NDA ReferencesPrivate NDA References15 TB credit card DW, 11 mil card holders15 TB credit card DW, 11 mil card holders

30 TB total storage, several TB+ instances30 TB total storage, several TB+ instances

1 TB Retail DW, relational query + complex OLAP1 TB Retail DW, relational query + complex OLAPAcorn profitability analysis PackageAcorn profitability analysis Package

1 TB relational instance with Complex OLAP1 TB relational instance with Complex OLAPLoss Prevention DW for store managementLoss Prevention DW for store management

Major RS operation on entire reporting tierMajor RS operation on entire reporting tierRuns on 8 dedicated reporting serversRuns on 8 dedicated reporting servers

Real-estate services, up to 10K report runs/dayReal-estate services, up to 10K report runs/dayMortgage closing: 3000 checks (Reports) /dayMortgage closing: 3000 checks (Reports) /day

Monitoring strategic venturesMonitoring strategic venturesVisual Studio .NET & .NET FrameworkVisual Studio .NET & .NET Framework

Page 8: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

How Do You Fit 22 Million How Do You Fit 22 Million Books Into 1 Data Books Into 1 Data Warehouse?Warehouse?

Faster access to Faster access to informationinformationDeeper view into key Deeper view into key performance performance indicators and trendsindicators and trendsBetter decisions for Better decisions for greater profitabilitygreater profitabilityImproved customer Improved customer experience experience

4.5 TB end-to-end 4.5 TB end-to-end data warehouse data warehouse running on SQL running on SQL Server 2005 64-bitServer 2005 64-bitStoring 3 years of Storing 3 years of transaction data; will transaction data; will grow to 5 yearsgrow to 5 yearsInsight enables better Insight enables better out-of-stock out-of-stock predictionspredictions

World’s largest World’s largest booksellerbookseller821 bookstores, 7.3 821 bookstores, 7.3 million retail itemsmillion retail itemsNeeded improved Needed improved business business intelligence for intelligence for merchandising and merchandising and inventory planninginventory planning

“SQL Server 2005 gives us the performance we need at a price that is just far superior to anything else we’ve seen.” Chris Troia, Chief Information Officer, Barnes & Noble

Page 9: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

Scenario/BusinessScenario/BusinessLargest bookseller in the USA, 850 stores, plus onlineLargest bookseller in the USA, 850 stores, plus online

40K employees, 1.5 mil titles across all stores, .75 mil at 40K employees, 1.5 mil titles across all stores, .75 mil at www.bn.comwww.bn.com

Retail Data WarehouseRetail Data WarehouseBroad use of SQL Server BI, DW & cubes started with SQL Server 2000Broad use of SQL Server BI, DW & cubes started with SQL Server 2000

Moved relational DW to SQL Server 2005 with June CTP nowMoved relational DW to SQL Server 2005 with June CTP nowFirst went live with SQL Server Integration Services in 2004 on Oct CTPFirst went live with SQL Server Integration Services in 2004 on Oct CTPLoading Data from mostly Oracle sourcesLoading Data from mostly Oracle sources

Workload/ArchitectureWorkload/Architecture2 TB sales & inventory data warehouse on SQL Server 20002 TB sales & inventory data warehouse on SQL Server 2000

Currently holds 3 years of transactions, will build up to 5 yearsCurrently holds 3 years of transactions, will build up to 5 yearsSSIS using Slowly Changing Dimension transform, lookups, and SSIS using Slowly Changing Dimension transform, lookups, and configurations configurations Will test Data Mining on IDW13 for a DM projectWill test Data Mining on IDW13 for a DM project

Out-of-stock prediction using Decision Trees, potential for enhanced sales of Out-of-stock prediction using Decision Trees, potential for enhanced sales of 150K/week150K/week

OLAP is planned to move pre-RTM:OLAP is planned to move pre-RTM:OLAP Subjects: Store sales, Store inventory, Distribution Center inventoryOLAP Subjects: Store sales, Store inventory, Distribution Center inventoryKey dimensions: Store, Item, Time, Demographic, Buyer, Vendor, CustomerKey dimensions: Store, Item, Time, Demographic, Buyer, Vendor, Customer

DB PlatformDB PlatformWindows Server 2003 64-bitWindows Server 2003 64-bitHP Superdome w/20 CPU partition for DW, 20 TB EMC SANHP Superdome w/20 CPU partition for DW, 20 TB EMC SAN

Barnes & Noble Retail DWBarnes & Noble Retail DWCustomer & Project ProfileCustomer & Project Profile

Live on SS2005!Live on SS2005!

Page 10: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

Barnes & Noble Retail DWBarnes & Noble Retail DWDW ArchitectureDW Architecture

OracleIMM

database

OracleSRS

database

Sales

ISStaging

database

Relationaldata

warehouseIS

Storeinventory

Distr.Center

inventoryStrategy

Inventory deltas

DimensionUpdates

POS transactions

Proclarityinteractive

reports

ReportingServicesreports

Inserts / Updates

Live on SS2005!Live on SS2005!

Page 11: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

First Premier Bankcard DWFirst Premier Bankcard DWTechnical Profile Technical Profile Scenario/BusinessScenario/Business

Specialized high-risk credit card issuer, 10th largest in the USASpecialized high-risk credit card issuer, 10th largest in the USA5 million card holder portfolio, 25K average new bookings per month5 million card holder portfolio, 25K average new bookings per month

Runs 100% of business on SQL ServerRuns 100% of business on SQL ServerGreat cost savings and business agilityGreat cost savings and business agilityLooking at SS2005 for entire BI stack: SSIS, Partitioning, SSRS & SSASLooking at SS2005 for entire BI stack: SSIS, Partitioning, SSRS & SSAS

Workload/ArchitectureWorkload/Architecture4.2 TB single instance for customer data4.2 TB single instance for customer data

Growth of approximately 300 GB per month, using DTS for data loadingGrowth of approximately 300 GB per month, using DTS for data loadingPlanning to upgrade this to SQL Server 2005Planning to upgrade this to SQL Server 2005

Ad hoc relational query workloadAd hoc relational query workload30-100 concurrent in-flight queries, 40 concurrent users at a tim30-100 concurrent in-flight queries, 40 concurrent users at a timee

Analysis Services OLAP usage through Proclarity & OutlooksoftAnalysis Services OLAP usage through Proclarity & Outlooksoft25-30 cubes, back-end DW is star-schema/snow-flake25-30 cubes, back-end DW is star-schema/snow-flake

Broad adoption of Reporting Services across the companyBroad adoption of Reporting Services across the company500+ unique reports, 1500+ report users500+ unique reports, 1500+ report users

Core 1.5 TB OLTP LOB system runs the businessCore 1.5 TB OLTP LOB system runs the business2500 customer reps connected all day, 200+ TPS frequently2500 customer reps connected all day, 200+ TPS frequentlyPlanning to upgrade this to SQL Server 2005Planning to upgrade this to SQL Server 2005

DB PlatformDB PlatformWindows Server 2003, SQL Server 2000 64-bitWindows Server 2003, SQL Server 2000 64-bitHP 64-bit 12-way, 8-way relational, 4-way OLAP, 27 TB EMC SANHP 64-bit 12-way, 8-way relational, 4-way OLAP, 27 TB EMC SAN

Page 12: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

First Premier Bankcard DWFirst Premier Bankcard DWData Flow Data Flow

Data is imported on a nightly basis from the different data sources and then standardized and validated through the Extract, Transform and Load (ETL) process. After the data has been validated, it is loaded into the data warehouse. It is then available to the Reporting and Analytic Departments for data analysis and development of reports.

Page 13: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

First Premier Bankcard DWFirst Premier Bankcard DWData Flows Data Flows

At the end of every month, FDR transfers a cardholder master database containing months of transactions conducted by customers of PBI to the PBI network, where the information is downloaded to the data warehouse. PBI uses SQL Server 2000 Reporting Services to create standardized company reports.

Page 14: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

First Premier Bankcard DWFirst Premier Bankcard DWChange ControlChange Control

The PBI data warehouse uses a 64-bit application with 16 GB of memory. The diagram layout emphasizes the testing and development environment more than the 64-bit technology.

Page 15: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

CROSSMARK Data WarehouseCROSSMARK Data WarehouseSystem ArchitectureSystem Architecture

Live on SS2005!Live on SS2005!

StagingWeekly

Transactions

OLE DB

Volumetrics ODS

52-week History

ReportsReport & Portal

Information

Microsoft Excel

HTTP

SQL Server 2005 Integration Services

TDLinx

AC Nielson Volumetric Data

(CSV)

Adobe Acrobat

HTML

Pre

sen

tati

on

Layer

Data

S

ou

rces

Tra

nsfo

rmati

on

Layer

Data

P

rocessin

g

Layer

SalesTrak ACNielsenICE

ACNielsen Volumetrics

Report Portal FrameworkReport Portal Framework

SQL Server 2005 Reporting Services

Page 16: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

Scenario/BusinessScenario/BusinessConsumer Packaged Goods (CPG) services provider and data processorConsumer Packaged Goods (CPG) services provider and data processor

16,000 employees, one of the largest CPG services companies in the USA16,000 employees, one of the largest CPG services companies in the USAEmploys in-store teams across the countryEmploys in-store teams across the country

Over 4 million store visits per year providing various inventory, stock and efficiency servicesOver 4 million store visits per year providing various inventory, stock and efficiency services

CROSSMARK retains top CPG manufacturers as customersCROSSMARK retains top CPG manufacturers as customersProcesses data on 60K UPC codes from sales in over 30K storesProcesses data on 60K UPC codes from sales in over 30K stores

Workload/ArchitectureWorkload/Architecture4 TB of data in a single SQL Server instance4 TB of data in a single SQL Server instance

25 TB of raw storage25 TB of raw storageUses Table Partitioning feature to increase manageability and reduce complexityUses Table Partitioning feature to increase manageability and reduce complexity

Ad hoc complex query workload throughAd hoc complex query workload through350 users (85 full-time power users)350 users (85 full-time power users)

SQL Server Integration Services for Data loadsSQL Server Integration Services for Data loadsData imported from 4 sources including, biweekly AC Nielsen data loadsData imported from 4 sources including, biweekly AC Nielsen data loadsPermitted expansion of DW from 13 weeks to 1 yr of historyPermitted expansion of DW from 13 weeks to 1 yr of history

SQL Server Reporting ServicesSQL Server Reporting ServicesDB PlatformDB PlatformWindows Server 2003, separate servers for ETL, Rel DW and ReportingWindows Server 2003, separate servers for ETL, Rel DW and ReportingHP ProLiant DL585 w/4 dual-core 2.4 GHz AMD 64-bit Opteron & 16 GB of HP ProLiant DL585 w/4 dual-core 2.4 GHz AMD 64-bit Opteron & 16 GB of RAMRAMHP StorageworksHP Storageworks

Live on SS2005!Live on SS2005!

CROSSMARK Data WarehouseCROSSMARK Data WarehouseSS2005 Solution ProfileSS2005 Solution Profile

Page 17: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

Talbots Retail Data Talbots Retail Data WarehouseWarehouseTechnical ProfileTechnical ProfileScenario/BusinessScenario/Business

Leading retailer & E-tailer with global catalog operationLeading retailer & E-tailer with global catalog operation1000+ stores in USA, Canada & UK1000+ stores in USA, Canada & UKWW catalog & online operation reaching 46 mil customers in 140 countriesWW catalog & online operation reaching 46 mil customers in 140 countries

Uses SQL Server as strategic database along with IBM mainframeUses SQL Server as strategic database along with IBM mainframeKey e-commerce and catalog order systems on SQL ServerKey e-commerce and catalog order systems on SQL Server

SQL Server 2005: Retail Data WarehouseSQL Server 2005: Retail Data WarehouseEstablished for business agility in retail operationEstablished for business agility in retail operationAssess store operation efficiencyAssess store operation efficiency

Workload/ArchitectureWorkload/Architecture1 TB of data, using SQL Server 2005 Table Partitioning feature1 TB of data, using SQL Server 2005 Table Partitioning feature

36-month of transaction history, partitioned using SQL Server by month36-month of transaction history, partitioned using SQL Server by month

Dual trickle plus batch data loading with SSISDual trickle plus batch data loading with SSISContinuous real-time trickle data, approx 1 million unaudited sale Continuous real-time trickle data, approx 1 million unaudited sale transactions/daytransactions/daySSIS batch loading of audited sales data, 1 million rows in 3 hour batch-windowSSIS batch loading of audited sales data, 1 million rows in 3 hour batch-window2 million rows total added to the database every day2 million rows total added to the database every day

Extensive usage of SS2005 Reporting ServicesExtensive usage of SS2005 Reporting ServicesParametrized reports from 50+ users hit the database all dayParametrized reports from 50+ users hit the database all day

DB PlatformDB PlatformWindows Server 2003, HP 64-bit ItaniumWindows Server 2003, HP 64-bit Itanium

Page 18: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

Clalit Health Patient DWClalit Health Patient DWSS2005 Solution ProfileSS2005 Solution Profile

Scenario/BusinessScenario/BusinessLargest HMO in Israel, 2nd largest WW, covers 60% of pop (3.7 mil)Largest HMO in Israel, 2nd largest WW, covers 60% of pop (3.7 mil)

14 hospitals, 1400 clinics, 6,000 doctors, 500 pharmacies, 32K employees14 hospitals, 1400 clinics, 6,000 doctors, 500 pharmacies, 32K employees

IBM Informix/Sun 10000 Solaris migration in 2002IBM Informix/Sun 10000 Solaris migration in 200267% TCO savings, 25% perf improvements: 15x in some OLAP queries67% TCO savings, 25% perf improvements: 15x in some OLAP queries

Also runs main clinic OLTP system on SQL ServerAlso runs main clinic OLTP system on SQL ServerMigrating 1000 local SQL Servers in clinics to central DB, 2 TB expectedMigrating 1000 local SQL Servers in clinics to central DB, 2 TB expected

Workload/ArchitectureWorkload/Architecture5 TB of total data, multiple servers & instances5 TB of total data, multiple servers & instances

2 TB in DW SQL Server instance2 TB in DW SQL Server instanceAd hoc query workload (Business Objects), 20 concurrent in-flight oftenAd hoc query workload (Business Objects), 20 concurrent in-flight often300 users (100 full-time) users, 10K queries/month300 users (100 full-time) users, 10K queries/month

OLAP cubes using Microsoft Analysis ServicesOLAP cubes using Microsoft Analysis Services3000 OLAP users, 45K queries/month ISV front-end app: Procalrity3000 OLAP users, 45K queries/month ISV front-end app: ProcalrityETL: Ascential Software’s Datastage ETL from S/390 mainframesETL: Ascential Software’s Datastage ETL from S/390 mainframes

Data Mining: Patient Treatment Cost PredictionData Mining: Patient Treatment Cost PredictionWorking with local DM specialist partner G-StatWorking with local DM specialist partner G-Stat

DB PlatformDB PlatformWindows Server 2003, Unisys 12-way, 64-bit, EMC SANWindows Server 2003, Unisys 12-way, 64-bit, EMC SAN

Page 19: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

SUN10000

Informix

ProductionDWH

SUN6000

Informix

StagingDWH

Sourc

e

Files

Files for

OLAP

Proclarity

LegacySystems

UnisysSQL

Server

Replicated DWH

for OLAP

OLAP

Unisys32-bitSQL

Server

StagingDWH

Unisys64-bitSQL

Server

Partitioned Unisys 64 bit

Unisys64-bitSQL

Server

ProductionDWH

Business Objects

Proclarity

OLAP

8 CPU32 GB1.5 TB

4 CPU16 GB

LegacySystems

Clalit Health Patient DWClalit Health Patient DWBefore & After Informix MigrationBefore & After Informix Migration

Before:

After:

Page 20: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

Customer Data:•Demographic data•Call Center•Mailing•Churn analysis•Complaints

Human Resources:•Manpower•Payroll•Outside contractors

Logistics:• Supplies• Purchases• Specialized medical equipment

Integrative Universes:•Clinic budget•Medical KPI’s•Medical services

Health Services & Expenses:• Medical services• Laboratory• Hospital services• Medication• Specialists• Medical imaging• Chronic illnesses• and more…

Clalit Health DW Clalit Health DW Subject Areas Subject Areas

Page 21: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

Scenario/BusinessScenario/BusinessUnited States Department of Veterans Affairs, Veteran’s Health United States Department of Veterans Affairs, Veteran’s Health AdministrationAdministration

172 Hospitals, 1260+ total points of care (including clinics & nursing homes)172 Hospitals, 1260+ total points of care (including clinics & nursing homes)5 million patients, $26 Billion budget, 200K employees5 million patients, $26 Billion budget, 200K employees

VHA Corporate Data WarehouseVHA Corporate Data WarehouseUses end-to-end Microsoft SQL Server and BI technologiesUses end-to-end Microsoft SQL Server and BI technologiesCurrently in the midst of 50 TB project to add new relational ODS and EDW back-Currently in the midst of 50 TB project to add new relational ODS and EDW back-endsends

Workload/ArchitectureWorkload/Architecture4 TB largest instance, 17 TB storage4 TB largest instance, 17 TB storageHeavy SQL Server Analysis, Reporting and Transformation Heavy SQL Server Analysis, Reporting and Transformation ServicesServices

100’s of cubes and subject areas100’s of cubes and subject areas

Main Data Marts: Lab (53 tests), Radiology, Pharmacy, ProstheticsMain Data Marts: Lab (53 tests), Radiology, Pharmacy, ProstheticsOutpatient: (Appointments, Encounters, Primary Care Panels)Outpatient: (Appointments, Encounters, Primary Care Panels)Inpatient: (Movement, Discharges)Inpatient: (Movement, Discharges)Other: Non VA Care, Human Resources, Financial AccountingOther: Non VA Care, Human Resources, Financial AccountingPlanned: Nursing, Dental, Purchasing, Health Data RepositoryPlanned: Nursing, Dental, Purchasing, Health Data Repository

Presentation Tier: ProClarity Analytics & Dashboard Servers, MS Presentation Tier: ProClarity Analytics & Dashboard Servers, MS MapPointMapPoint

DB PlatformDB PlatformMS Windows Server 2003 Enterprise EditionMS Windows Server 2003 Enterprise Edition

4 x HP RX5670 64-bit 4-way Itianium-2 Servers, 32 GB RAM, HP EVA SAN4 x HP RX5670 64-bit 4-way Itianium-2 Servers, 32 GB RAM, HP EVA SAN

US US Veteran’s Health AdminVeteran’s Health Admin CDWCDWTechnical ProfileTechnical Profile

Page 22: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

VHA Data Warehousing FrameworkVHA Data Warehousing Framework

WaitWaitTimesTimes

DataDataWarehouseWarehouseVHAVHAafaf

VHAVHAcc

SourceSourceSystemsSystems

DiabetesDiabetes

AcquireAcquire Populate Populate Create Create Access Access Data Data WarehouseWarehouse MartsMarts Information Information

11 33 44

CommonCommonQuery, Reporting,Query, Reporting,

Analysis, and Analysis, and Data MiningData Mining

ToolsTools

OPOP

PBMPBM

22

OtherOther

VistAVistAHDRHDR

ADRADRABCABCG/LG/L

DoDDoDCDCCDCCMSCMS

ConformedConformedDimensionsDimensions

Prog OfficeProg OfficeData MartsData Marts

Ext

ract

, Tra

nsf

orm

, Lo

adE

xtra

ct, T

ran

sfo

rm, L

oad

VISNVISNWarehousesWarehouses

VHAVHAcc – VHA clinical systems – VHA clinical systemsVHAVHAafaf – VHA administrative & financial systems – VHA administrative & financial systems

Program OfficesProgram Offices•Pharmacy BenefitsPharmacy Benefits•ProstheticsProsthetics•DentalDental

Closed Loop Information SystemClosed Loop Information System

DataDataConsultantsConsultants

ResearchResearchData MartsData Marts

Value Added DataValue Added Data

MetadataMetadataRepositoryRepository

Page 23: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

DimCPT

CPTID varchar(5)

CPTName varchar(30)CPTCategory varchar(75)CPTClass varchar(40)Inactive varchar(8)WorkRVU numeric(10,2)PrExpRVU numeric(10,2)MalPrRVU numeric(10,2)TotalRVU numeric(10,2)

DimDRG

DRGID smallint

DRGName varchar(70)DRGCategory varchar(70)DRGClass varchar(7)DRGType char(3)ValidYear int

DimICD9

ICD9ID int

ICD9Txt varchar(9)ICD9Prefix varchar(5)ICD9Name varchar(50)ICD9Category varchar(50)ICD9Class varchar(100)ICD9SubClass varchar(100)

Clinical

DimAge

AgeID smallint

AgeGrp2 varchar(15)AgeGrp4 varchar(15)AgeGrp10 varchar(15)AgeGrp14 varchar(15)AgeGrpARC varchar(15)

DimPatient

ScrNumID int

ScrSSN varchar(9)PtSSN varchar(9)ICN char(10)DoB smalldatetimeDoD smalldatetimeDoDSource varchar(10)Gender varchar(1)Race varchar(25)EnrZip intHmFIPS intPreferredVISN intPreferedSta3n intPreferredSta6a varchar(6)CFYVERAClass char(3)PFYVERAClass char(3)Eligibility char(3)EnrPriority varchar(3)EnrStatus varchar(50)PCProvider char(10)Diabetes char(1)Hypertensive char(1)

Demographic

DimARCTrtLoc

TrtLocID char(3)

TrtLocName varchar(100)TrtLocCategory varchar(100)

DimARCTrtType

TrtCodeID varchar(5)

TrtCodeName varchar(50)

DimALBAcct

AccountID smallint

AcctDesc varchar(50)

DimCostCtr

CostCtrID smallint

CCName varchar(100)CCCategory varchar(50)CCClass varchar(50)

DimALBProductionUnit

ProdUnitID char(2)

PUName varchar(125)PUCategory varchar(50)PUClass varchar(50)

DimARCDxClass

DXClassID smallint

DXClassName varchar(50)DXClassARC nvarchar(2)

DimARCEligibility

EligibilityID char(2)

EligibilityName varchar(100)EligibilityCategory varchar(100)

DimARCIncome

IncomeID smallint

IncomeGroup varchar(50)

Financial

DimBOC

BOCID smallint

BOCName varchar(50)BOCCategory varchar(50)BOCClass varchar(50)

Geographic

DimVISN

VISNID smallint

VISNTxt char(3)VISNName varchar(50)City varchar(75)State char(2)Zip intFIPS intFTE smallintBudget money

DimFacility3n

Sta3nID smallint

Sta3nName varchar(100)VISN smallintCity varchar(50)State varchar(2)Zip intFIPS intMCG smallintFTE smallint

DimFacility6a

Sta6aID varchar(6)

Sta6aName varchar(100)Sta6aType varchar(50)Sta3n smallintVISN smallintVASTID smallintSta6aCity varchar(100)Sta6aState char(2)Sta6aZip intSta6aFIPS intAdminParentCode intAdminParentCity varchar(100)AdminParentState char(2)AdminMCG smallint

DimFIPS

FIPSID int

State char(2)County varchar(30)VISN smallintVISNCARES smallintCARESMkt varchar(100)VetPop int

DimZip

ZipID int

State char(2)City varchar(100)

Hospital Location

DimStop

StopID smallint

StopName varchar(50)Category varchar(50)Class varchar(50)Type char(10)

DimBedsection

BedsectionID smallint

BedsectionName varchar(50)Category varchar(50)Class varchar(50)Type char(10)

Time

DimDate

DateID int

MDate smalldatetimeFYr char(4)FQtr char(6)CMth char(5)CDay varchar(30)FP smallintDoW char(3)DoM smallint

StandardizedType/Size conventionsNaming conventions

Verified“Gold” standardBusiness rules

OptimizedPrimary keysIndexed

Refreshed

VHA Dimensional MapVHA Dimensional Map

Page 24: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

VHA CDW Hardware VHA CDW Hardware ArchitectureArchitecture

FTP, Import,Staging

(4x) 32-Bit Server

MasterDatabase

Cube Dev

Cube R&D

Cube Prod

Cube Dev

(4x) 64-Bit Server

Web Server

(2x) 32-Bit Server

Power User

Executive User

General User

Yukon R&D

Report Services

VISNData

Warehouse

Gigabit Ethernet

Delimited&

Fixed LengthText Files

(4x) 32-Bit R&D Server

Page 25: SQL 2005 Customer Evidence Contact: Ram Ramanathan.

ETLSubsystem

DatabaseSubsystem

StorageSubsystem

(HP EVA 8000 - 24TB)

ETL1 - DL385 - x64DC - 4x

ETL2 - DL385 - x64DC - 4x

ETL3 - DL385 - x64DC - 4x

42U Cabinet 42U Cabinet 42U Cabinet

ETL4 - DL585 - x64DC - 8x

ETL5 - DL585 - x64DC - 8x

DB1 - RX7620 - IA64 - 8x

DB2 - RX7620 - IA64 - 8x

HP ProCurve 3400cl Gb Switch

Clu

ster

ed

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

M5314A Enclosure - 14x -142GB - 15k RPM

HSV210 FC Controller

HSV210 FC Controller

MSL6060 - 2

AA

C F

irew

all

NightlyUpdates

LargeBatches

ETL Server Type 1: HP DL385, AMD Opteron Dual Core, 2 cores/4 processors, 16GB, (6) 72GB Drives, (2) Gb NICs, no SAN HBA

ETL Server Type 2: HP DL585, AMD Opteron Dual Core, 4 cores/8 processors, 16GB, (4) 72GB Drives, (2) Gb NICs, (2) SAN HBA’s

DB Server Type 1: HP RX7620, Intel Itanium-2, 8 processors, 32GB, (4) 72GB Drives, (2) Gb NICs, (2) SAN HBA’s, Clustered

Server Legend:

VHA Corporate Data Warehouse System(Primary Function: ETL Import and Export, Data Warehouse Storage)

Streams

Batches

ETL4 - DL385 - x64DC - 4x

SAN Switch(2Gb FC)

64-bit: Windows, SSIS, SSDE

64-bit: Windows, SSIS, SSDE

64-bit: Windows, SSDESSIS - SQL Server Integration ServicesSSDE - SQL Server Database Engine

Future Expansion Future Expansion

Dual Core

•Discussed many scenariosand obtained manyconfigurations.•64-bit end-to-end

Page 26: SQL 2005 Customer Evidence Contact: Ram Ramanathan.