re: Using DBXL with Existing Databases

Patrick Halstead — Wed, 28 Apr 2010 21:57:52 GMT

Update: Hilary Stoupa wrote a blog with examples of how to use DBXL on existing SQL databases: www.infopathdev.com/.../use-dbxl-submit-with-existing-sql-data.aspx.

While we update top-level tables in place, we choose to do cascading deletes for child and grandchild (etc.) tables because an update would require a tight coupling and the whole point of the DBXL Web Service is to provide a loose coupling. Since InfoPath does the data type and structure validation for you, there’s no need to tightly couple at the submit interface anymore so long as you have InfoPath. Loose coupling enables greater up-time because the mapping can fail but since DBXL always stores the XML, the admin can go in after the fact and re-shred the data. Loose coupling also enables data source changes and many-to-one mapping where separate XMLs are mapped to the same table set (a boon for versioning). We always want the submit to succeed. Reads can sometimes fail, which is ok, but if submits always succeed then it provides a more robust and scalable system for business process automation.