astrazeneca data migration project - sascommunity · Ł map legacy and current data using wa!final...

39
AstraZeneca Data Migration Project Dave Smith - SAS UK Paul Frost - AstraZeneca Copyright ' 2000 , SAS Institute Inc. All rights reserved.

Upload: lamhuong

Post on 04-Jun-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

AstraZeneca Data Migration Project

Dave Smith - SAS UKPaul Frost - AstraZeneca

Copyright © 2000 , SAS Institute Inc. All rights reserved.

Agenda

� Project description� Methodology� Technical Architecture� Consultancy Deliverables� Project Progress/Status� Conclusions

Project Description� Migration of data from two data entry systems

into a common format � One system will eventually be

decommissioned� Data needs to integrate with other systems

(e.g. reporting and validation programs)� Process needs to support compliance with

Pharmaceutical regulation 21CFRPart11

Legacy database system� Running on a CMS platform� FOCUS based system

!Hierarchical system!Difficult to query!Standard extract processes => SAS based on

Cobol!Standard pre-processing programs available to

re-format data

Current database system� In-house developed system (ex-Astra)

running on Sun Solaris� Oracle based� VB GUI� Standard extract processes => SAS� Standard macros available to convert SAS

output to generic data standards

Major System differences� Hierarchical versus relational� Oracle DB is highly normalised

!Loading data into Oracle would require extensive re-formatting

� Many studies have pre-written reports based on legacy structure!We need to maintain this functionality

� New RDBMS has associated validation toolset!We need to be able to utilise this

Migration options� Extract legacy data, re-format and load into

Oracle!High resource overhead and specialised skill sets

required� Re-key data into new RDBMS

!Not viable for large studies� Create copy of existing database in new

RDBMS (as far as possible)!Two disparate data sources!Integrate within a data warehouse

Preferred strategy� Extract legacy data using pre-defined processes

!Create static, R/O copy of data to act as source data� Create �copy� of legacy database in new

RDBMS� Map legacy and current data using WA

!Final integrated data needs to look like the legacy data but contain additional variables to allow use of new validation tools

� Use new validation tools and legacy report programs against integrated data

Chosen Methodology

� SAS/Warehouse Administrator® software for control of the process!Off the shelf product from strategic vendor!Support through SAS version upgrades

� SAS Intelligence Architecture to facilitate deployment

SAS/Warehouse Administrator®

� Single point of control � Manages complex data integration

processes� Manages metadata throughout the process

! Impact Analysis ! Audit Trails

� Self-documenting

What�s behind theSAS Intelligence Architecture?

FMSFMS

SolutionSolution AdaptorsAdaptors

CRMCRMSRMSRM

EPMEPM

ApplicationIntegrationPlug-ins

DataIntegration

Plug-ins

CaptureCapture

TransformTransform

IntegrateIntegrate

ExtractionExtraction AdaptorsAdaptors

Suppli

ers

Suppli

ers

Organ

isatio

n

Organ

isatio

n

Custo

mers

Custo

mers

EnterpriseEnterprise

Summar

y

Summar

y

Foun

datio

n

Foun

datio

n

Detail

Detail

Staging

Staging

Snaps

hot

Snaps

hot

ODDODD

RawRaw

CaptureCapture TransformTransformQQ IntegrateIntegrate ExploitExploit

Solu

tion

Solu

tion

Ada

ptor

sA

dapt

ors

Extr

act

Extr

act

Cle

anC

lean

Valid

ate

Valid

ate

Load

Load

Join

Join

ExceptionsExceptions

Tran

sfor

mTr

ansf

orm

Der

ive

Der

ive

Extr

actio

nEx

trac

tion

Ada

ptor

sA

dapt

ors

ERPERP

EAIEAI

VSAMVSAM

FTPFTP

RDBMSRDBMS

Intelligence Architecture Blueprint

OperationalOperational Intelligence ArchitectureIntelligence Architecture Business Business SolutionsSolutions

Where this helps AZ�� Consists of

!Best Practices!Consistent Structure!Extra Applications

� IAB Configuration Manager� IAB Job Administrator

� Leads to:!Easier maintainability!Faster deployment!Easier Scheduling of jobs!More IT rigour

Technical Architecture

!SAS/Warehouse Administrator®

!SAS Intelligence Architecture (SASIA)!Bespoke SAS/AF® Applications

Oracle Server

SAS ServerSAS Temporary Work tables

MainframeMetadata ServerWarehouse Metadata,

Audit Editing Application

WA ClientWA Client

Data serverWarehouse and Raw data

Warehouse Environment

Physical Architecture

Legacydata

Currentdata

Editable Layer

OperationalEnvironment Warehouse Environment

Business & Technical Metadata

Raw Layer Staging Layer Integrated Layer

Data Warehouse Architecture

Warehouse structure Screenshot

Process Map Screenshot

Integration Layer Screenshot

Job view Screenshot

Job Admin Screenshot

LSF v Job Administrator

� Did not use LSF because!No need for load balancing between processors!Simple scheduling requirements

� Can use the two together later if required; publish in job administrator and submit with LSF

Job admin old screenshot

IAB Configuration Manager

� Part of IAB toolset� Makes it easier to move from Development

to Production environments� Consistent structures are easier to replicate

between versions of the warehouse� Also aids replication of structures between

studies

Consultancy Deliverables

� IAB Infrastructure� First Warehouse Build� AF Applications

!Audit Data Changes!Manage Security

� Training

IAB Infrastructure

� Deploy Test study� Check Connectivity

!Servers!Connect Spawner!Share Server

� Add Auditing Applications

Using the SAS Intelligence Architecture

Test ProjectTest

Project

Project 5Project 5

Project 1Project 1

Project 3Project 3Project 4Project 4

Intelligence ArchitectureIntelligence Architecture

Project 2Project 2

INTEGRATED EXTENDABLE ARCHITECTURE

FOCUSED ON BUSINESS ISSUES

BASED ON EXPERIENCE

First Warehouse Build

� Gather Requirements� Define Architecture� Create First Process� Publish to Job Administrator

Auditing Applications

� Built in SAS/AF®

� Needed to ensure that the editable layer is managed in a manner which can be compliant with 21CFRpart11!Any data alteration must be tied to the user

identifier and a date-time stamp!Access to alter the data must be controlled!Updates must be audited with a reason for

change

Auditing Applications

� Each dataset has a password applied using a compiled macro

� The macro resolves to the password depending upon the user identifier

� The macro is compiled by the SAS/AF®

application administration screen� Users and administrators never see the

password

Auditing Applications

� Editing of datasets is via a separate SAS/AF® subclass, which !Checks that the audit trail has been

invoked !Forces the user to enter a reason for

change� This application screen is associated as the

Open command within the warehouse

Editing Datasets

Audit Trail

Training

� Train users in WA� Skills transfer of IAB� Overview of Rapid Warehousing

Methodology� Skills transfer in Auditing Applications

Demo

Project Progress/Status

� SAS Consultancy used to assist with Warehouse implementation and best practices

� Latest versions of IAB implemented� Multiple projects successfully migrated� Also used to manage long-term safety data

Conclusions

� SAS/Warehouse Administrator® and SAS Intelligence Architecture have been used to create a controlled process

� This gives reassurance to both AstraZeneca and the regulatory bodies

� Users had never seen WA before, although they were extremely SAS literate

� From first meeting to first study build in WA took three months

� SAS Consultancy only 19 days � on time & budget!

Questions