You can read the archive / “History” directly with a Microsoft SQL Server Spatial or Non-Spatial Reader.

They will be the same table name as the base table, but with “_H” appended.  You can see the table name in ArcCatalog

 

​@bwn unversioned featureclasses have the archive in the same table, it has GDB_FROM_DATE and GDB_TO_DATE that delimit when a feature has been valid. Reading archive history one can do in an ArcSDE reader (using archive where clause) or any MSSQL reader.

Workflow I have so far:

• Export xml workspace document without data (esri tool) for featureclass I want to migrate
• Import xml workspace into new egdb (also ESRI tool)
• Then one needs to update the featureclasses in the new egdb to set archiving enabled for the target featureclasses (that have archiving in the source)
• A simple read/write FME operation using MSSQL Spatial reader/writer
  • feature classes first
  • relational tables second
• This seems to work fine, except that the OBJECTID’s reference for new features in the egdb will be out of sync, but it seems to doable to check out new OBJECTIDs using the esri stored procedures to work around it

(haven’t looked at attachments yet)