Large, infrastructure-based organizations such as utility companies often find that they need to operate with multiple GIS and CAD systems as part of their computing environments. This can create numerous challenges with data management and sharing of information within the organization. These issues can be addressed in part by the use of Spatial Extract-Transform-Load (ETL) technologies working in an enterprise environment along with a common data repository. This presentation highlights experiences SBS has had in using Spatial ETL technologies in organizations to provide an advanced architecture for sharing spatial data. There are a number of technical considerations that must be addressed in these environments. These include: • Support for multiple GIS platforms that all access a common data base • A common, canonical data model that supports existing data models in the enterprise • Cross platform conflict resolution • A common data validation framework • Global ID management • A comprehensive network model to support business application requirements • Scalability to support a very large user community • Integration with other systems to support business workflows and advanced analytics

1. Using Spatial ETL in a Multi-Vendor Enterprise
GIS Environment
Dennis Beck, PE
President

2. Agenda
•
•
•
•
Introduction
Multi-vendor architecture problem statement
Challenges in implementing multi-vendor solutions
Case study architectures
Proprietary and Confidential

3. About Spatial Business Systems
Delivering the power of spatial information
•
Full-services solution provider for geospatial technologies in the
utility, telecommunications and government market sectors
• Integrators of major spatial tools and platforms
•
Offices in Lakewood, CO and Melbourne, Australia
•
Data management, project management and spatial application
business strategy consulting
•
Spatial integration product solutions
Proprietary and Confidential

4. Some of Our Clients
Proprietary and Confidential

5. Problem Statement (1)
•
Infrastructure-based organizations have tried to consolidate multiple
spatial/graphical platforms that exist across different business functions:
–
–
–
–
–
–
•
•
Design
Planning
Construction
Operations
Field users
“GIS”
The problem has typically been addressed via platform consolidation efforts
Multiple systems still proliferate within these organizations
Proprietary and Confidential

6. Problem Statement (2)
•
There are multiple reasons why these systems proliferate
–
–
–
–
–
Mergers and acquisitions
Legacy systems that are hard to migrate
Specialty, packaged spatial applications
Unique requirements by different user communities
Prevalence of CAD
•
The problem is continuing, rather than going away
•
This is creating new architectures that are focused on
– Data management
– Interoperability
– Well-defined integration points and business controls to ensure data integrity
Proprietary and Confidential

7. Challenges in a Multi-Vendor GIS Environment
•
GIS vendors typically support spatial data management functions
within their product suites
–
–
–
–
–
•
•
Unique modeling requirements
Unique approach to long transactions / conflict resolution
Unique indexing approaches
Unique access methods
Unique formats
Complexity – more systems to manage
Operational issues
– Training
– Support
– Change management
•
Data integrity
Proprietary and Confidential

8. Approaches
• Loose coexistence (Business as usual)
• GIS platform as hub (Geo-Centric)
• Database as hub (Database-Centric)
Proprietary and Confidential

9. Geo-Centric Architecture
•
•
•
The GIS platform is the source of truth
External products connect to the GIS
The GIS is responsible for data management functions
– Validation
– Vendor’s data model, network model, metadata structures
– Conflict resolution
• Example – CAD / GIS integration
– Loosely coupled approach: ETL (Extract-Transform-Load)
– Tightly coupled approach: dynamic integration, e.g. OSGeo Feature
Data Objects (FDO)
Proprietary and Confidential

10. Geo-Centric – FDO Use Case
VanBuren
Natively access
data from multiple
spatial sources
Avoids
Runner
Jetson
Berry
Tomato
Franklin
Banana
 Conversion
 Data loss
 Data copies
 Stale data
Lang
Warren
Campbell`
Apple
Lincoln
Peach
Regan
Carter
Bradshaw
Pear
Clinton
Jackson
Grape
Jefferson
Carrot
Beet
Orange
Plum
Fraiser
Wilson
Adams
Kennedy
Roosevelt
Miller
Armey Baker Diego
Harvard
Ford
Varney
Jones Smith
Zoe
Washington Truman
Yale
Dartmouth
Cornell
Merrit
Eisenhower
Parcel data from
a SHP file
Utility data from
Smallworld VMDS
Property data
from Microsoft®
SQL Server™
Zoning data
ESRI® ArcSDE®
Aerial photos
Source: Autodesk

11. Database-Centric Architecture
•
Database is the source of truth
– Not the GIS vendor’s application database
– Typically a spatially enabled relational database (Oracle, SQL)
•
•
•
GIS platforms serve application-specific roles
Spatial ETL exchanging data between systems
The spatial database environment takes on an added role in
managing spatial data
–
–
–
–
–
•
Long transactions
Conflict resolutions
Metadata management
Symbology
Additional spatial capabilities – e.g. network modeling
Examples: Large investor-owned utilities
Proprietary and Confidential

12. High Level Data Architecture
DMS
Smallworld
Application
Database
Staging
Canonical
DataStore
Map3D
Application
Database
Landbase
Landbase
Updates
Updates
Operational
DataStore
Canonical
DataStore
Mirror
Spatial
Data
Warehouse
DataMart DataMart DataMart DataMart
Staging
Staging
Staging
AUD
Web
12

13. Operational (Spatial) Data Store Architecture

14. Use of Spatial ETL (Extract-Transform-Load)
•
•
•
•
•
•
•
•
Data model transformation support
– Attribute modification / Feature merging and splitting
– Relationship generation / Topology generation
Validation
Error handling with notifications via email, logs, tweets, NT Service logs
Can support full synchronization with appropriate plug-ins
Scalability via the use of server-based approaches (e.g. FME Server)
COTS support, particularly for upgrades
Web based initiation, interaction and resolution of synchronization
processes
Flexibility to support current and future format requirements
14

15. ETL Example: Bi-Directional Process Flow
Example: GE Smallworld and Oracle Spatial
Role of plug-ins and transformers
•
•
•
•
•
Format support
Model transformations
Difference management
Common (Global) ID management
Network model extraction

16. Conflict Resolution
•
As data is merged and posted between workspaces, the potential for
conflicts in the long transaction data exists
–
–
•
Mechanisms must be in place to detect changes
–
–
–
•
•
•
Such situations occur when two users modify the same feature at the same time, but in
different workspaces.
In this case merge / post operations may result in overwriting one user’s data in favor of the
other
Oracle Workspace Manager provides tools for comparing changes from different workspaces
and can identify sources of data change conflict
Change-Data-Capture (CDC) offer alternate mechanisms to identify differences
Trigger-based approaches are also used
It is necessary to visualize the changes to support comparison and
resolution of changes via semi-automated or manual processes.
Conflicts can be identified and resolved in the system, resolution results are
sent to the peer
Properly designed business processes will help to mitigate most conflicts
Proprietary and Confidential

17. Global ID Model
•
GUID_REGISTRY
– Repository / Mapping for each instantiated GUID
– ODS Feature: ODS object where this object acts as a Primary Key
– Source Feature: Identifier for the specific feature this data is brought over from,
in the source system.
– Source System: System where this data existed previously
– Key: Unique Identifier for this record in the Source System, for the Source
Feature object (These 3 values combine with the specific ODS Feature to form
a Unique Index on GUID_REG)
– The Global ID API is designed to leverage stored procedures in the ODS
wrapped with ETL transformers
17

18. Other Considerations
•
Common validation framework
–
–
•
Valuable for ensuring consistent representation
Source system validation may not address all use cases
Common symbology
–
–
–
Multiple symbol sets from source applications
Necessary to persist some in the ODS
Symbol and Text Stroking used for Web / Thin client deployment
•
•
–
•
Display scales
Common network models in the data base
–
–
•
Converts point and text to multi-linestring with symbol/font/size/orientation embedded
Implemented via ETL Transformers
Auto-generated via ETL plug-ins
Supports externalized application requirements
Application services
Proprietary and Confidential

19. Multi-Vendor Architectures – Value Proposition
Why bother?
•
Productivity improvements
– Use the best tool for the job, e.g. CAD
– Specialized applications to support trained staff members
•
•
•
•
Eliminate redundant data entry
Reduce custom developed applications
Reduced upgrade costs via model transformation
ODS as an integration hub
–
–
•
Value of improved data management
–
–
•
•
Lower cost integration
Improved maintainability
Currency of data
Availability to broader user base
Enables more advanced mobile computing
Support phased platform migrations
Proprietary and Confidential

20. Questions ?