Upload
calvin-todd
View
222
Download
2
Embed Size (px)
DESCRIPTION
3 Data Mastering Approaches Considered Custom code –Expensive to build and maintain –Performs poorly with unpredictable/unstructured data –Difficult/expensive to maintain Traditional software –Performs poorly with unpredictable/unstructured data –Relies on custom code ‘extensions’ Manual effort –Expensive –Non-scalable –One-time fix –Inconsistent result Custom code Semantic- based Data Quality (Oracle Product Data Quality) –Scalable, consistent results –Designed for non-standard data in many categories –Actually works! DI DQ
Citation preview
Emerson – Driving Data Emerson – Driving Data Standards Enterprise-Standards Enterprise-Wide Wide
Phil LoveManager, Data QualityLiebert Corporation
2
Emerson Corporation Emerson Corporation
$20B diversified global manufacturer
Growth through acquisition
Historically autonomous operations– 22 Divisions– 40 Systems– 268 Locations
Heterogeneous IT landscape
Siloed operations
Good News: Across the board migration to
Oracle Applications
Bad News: Data inconsistency will make the
transition difficult
The Answer: Single-instance MDM (eventually) Enforce standards across the
business (NOW!)
3
Data Mastering Approaches ConsideredData Mastering Approaches Considered
Customcode
– Expensive to build and maintain– Performs poorly with
unpredictable/unstructured data– Difficult/expensive to maintain
Traditional software
– Performs poorly with unpredictable/unstructured data
– Relies on custom code ‘extensions’
Manualeffort
– Expensive– Non-scalable– One-time fix– Inconsistent result
Custom code
Semantic-based Data Quality (Oracle Product Data Quality)
– Scalable, consistent results– Designed for non-standard data
in many categories– Actually works!
DIDQ
4
Why is the data so poor?Why is the data so poor?
Enterprise Data
Enterprise Standards
• Data spread across many systems
• Many differing objectives• Much good-faith effort,
but no consistency or scalability
• Inconsistent• Missing information• Different formats • No standards or different
standards
?No practical way for standards to be •Agreed •Coordinated•Enforced
5
The Missing Link: Standards EnforcementThe Missing Link: Standards Enforcement
Enterprise Data
Enterprise Standards
• Standards are built into Data Mastering Process
• Constant feedback improves the standards
• Data is evaluated against standards
• Immediate integrated remediation, as required
• Custom publishing for systems that need data in a different form
Enforcement is a virtuous cycleStandards grow and adapt based on real-world usage
Data standards are maintained and enforced by the DataLens System Data Mastering services
6
Data Mastering BenefitsData Mastering Benefits
Automation of– Classification
• For procurement (UNSPSC, DRI)
• Import/export regulations (HTS)
– Attribute standardization
– Description standardization
– Enrichment– Validation
Increases the value of our data!
Drives quality & consistency
Reduces lag timeReduces cost
Prepares the way forSystem migrationOther forms of MDM
7
Data Mastering – Phased Rollout Example Uses Data Mastering – Phased Rollout Example Uses
Phase 1 – Build the engine– PLM clean-up – enrich, standardize, identify duplicates
Phase 2 – Expand across the Division– Interim Item hub – cleanse, de-dup & load– International divisions – translation, systems cut-
over– Migrate to Oracle Apps – load, standardize, validate – Expand coverage – to Assemblies & Finished goods
Phase 3 – Expand across Divisions/Enterprise – Corporate material catalog – standardize and validate – Data Warehouse – standardize, classify for reporting– Procurement – standardize classifications and feeds – Partner Portal – translate languages, optimize for search– Pricing system – standardize & validate load data
Productivity: Broad automation allows focus on exceptions
Leverage: Ultimately, Data Mastering will touch ~75% of enterprise systems
8
Standardize & validate for load
ItemHub
Search
X-RefExternal
DB
Cleanup legacy
Transform and integrate between systems
Drive Standards System-by-System, Drive Standards System-by-System, Process-by-ProcessProcess-by-Process
9
Drive Standards System-by-System, Drive Standards System-by-System, Process-by-ProcessProcess-by-Process
Enterprise Data Mastering•Single place to maintain all standards
•Single place to enforce all standards
10
Drive Standards Division-by-DivisionDrive Standards Division-by-Division
Division 1
Division 4
Division 3
Division 2
11
Lessons Learned: Lessons Learned: Governance, Standardization and MDMGovernance, Standardization and MDM
Think Big – start small
‘Traditional’ approaches won’t work – not generalizable or scalable
Semantic-based Data Mastering with the DataLens System– Delivers rapid tactical benefits– Allows for phased rollout – Avoids traditional data management ‘gotchas’
Necessary starting point for any MDM strategy– Data Standardization– Data Remediation– Data Governance
– Development and Enforcement of Standards!