It would be terrific to have a UML Profile of the Geodatabase for Geodatabase Design. Then one could use the UML tool of one's choice for Geodatabase Design (Enterprise Architect, etc) and not be locked into using Visio. Diagrammer is great, but it's an unsupported research project; and it can't be used to trace GDB items back to requirements and use cases.
Since the XML Schema of the Geodatabase can exhaustively describe everything in a geodatabase, a UML Profile of the Geodatabase should be possible in principle. Translators / translation code may be an issue. But since ArcCatalog can import and export XML Workspace Documents, I don't see why a UML Profile couldn't be developed that, if used, could generate valid XML Workspace Documents and thus be loaded into ArcCatalog for geodatabase instantiation. We really need something better for geodatabase design than what we presently have.
It may ultimately be up to the user community to develop a UML Profile of the Geodatabase. UML may be a bit too obscure / underadopted by the geodatabase design community for Esri to commit the resources to this. However, I agree with commentators in other places who bemoan the inability to use good CASE tools such as those that exist for relational database design--and the difficulty in communicating geodatabases with the wider community of non-GIS IT folks like DBAs, Enterprise Architects, etc. A UML Profile of the Geodatabase, along with a way to provide full round-trip design, would go a long way toward integrating GIS into the broader bucket of Enterprise IT and beyond. GIS is growing up. Would love to see geodatabase design mature in step with this trend.
A) The key workflow must include all GDB components such as domains. Using a single GDB that has one of every data type and field, would be a great way to validate this tool is working correctly.
B) Use ability defines adoption. If the tool is made, please make sure it handles the extremely complex configuration issues that other ESRI tools seem to miss their first time out.
- Handle relationship loops.
- Handle multiple features with the same alias name.
- Handle the export and import of domains, even if they do not appear assigned to a feature class. This is constantly overlooked.
- Provide extremely simple logic to pull out the Schema, avoid putting in erroneous logical checks that are not documented or enforced anywhere else.
- Fix the XML export so it also follows this logic.
I'm helping organize a Special Interest Group at the 2011 Int'l UC to facilitate discussion about GDB modeling:
Geodatabase modeling and design SIG and survey at 2011 Esri Int'l UC
Description from UC agenda:
http://events.esri.com/uc/2011/infoWeb/OnlineAgenda/?fa=ofg_details_form&ScheduleID=2755
“This Special Interest Group will discuss available and forthcoming GDB design tools, patterns, and emerging issues. The group will provide participants with an opportunity to communicate their database design and modeling requirements to Esri staff who will be present.”
Here's the link to the online agenda:
http://events.esri.com/uc/2011/infoWeb/OnlineAgenda/?fa=ofg_details_form&ScheduleID=2755
For more information visit: http://spatialdba.com
Please take a minute to fill out the following survey:
http://spatialdba.com/limesurvey/index.php?sid=72916〈=en
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.