使用新架构更新XML列架构

Question

I am using SQL Server 2008 and busy doing a POC where I have a table with a typed XML column. The nature of the POC is that the XML will change over time as our needs evolve for the data that gets stored in it. (We are serializing objects in C# to persist the data to the DB)

However, I need to keep the old XML in the table for auditing purposes and be able to validate any new XML that gets added to it correctly with the updated schema.

I have tried dropping the schema from the collection, updating it (in this case adding a required attribute), adding it back to the collection and then binding it to the column it belongs to. When doing so I simply get a validation error that the XML already in the column fails the validation because it does not contain the new attribute.

Is there a way to update the XML column to use a new Schema and ignore the XML that is already stored in the column? Or would it be possible to disable validation while adding the new schema to the XML column?

Answer 1

A schema collection should permit you to have different versions of the XML data without having to ignore validation.

Have a look at XML Schema Collections to start with.

In your T-SQL code, you'd have something like this (this code is sourced from Bob Beauchemin of SQLskills.com):

-- Load XML schema from file
DECLARE @x XML
SET @x = (
SELECT * FROM OPENROWSET(
   BULK 'C:\invoice.xsd',
           SINGLE_BLOB
   ) AS x
)

-- And use it to create an XML schema collection
CREATE XML SCHEMA COLLECTION InvoiceType AS @x

Now you create a table which maps the column to the schema collection:

CREATE TABLE invoice_docs (
   invoiceid INTEGER PRIMARY KEY IDENTITY,
   invoice   XML(document InvoiceType)
)

Now when your schema changes, you modify the schema collection by adding in the new version of the schema:

DECLARE @x XML
SET @x = (
SELECT * FROM OPENROWSET(
   BULK 'C:\invoice_v2.xsd',
           SINGLE_BLOB
   ) AS x
)

-- And use it to create an XML schema collection
-- Allow V1 and V2 invoices
ALTER XML SCHEMA COLLECTION InvoiceType ADD @x

The old data already in the table is validated against the old schema, and any new data will validate against the old or new schema.

If you want to validate only against the newer schema, you'll have to add an additional constraint to the column.