Thanks, perhaps a bit more background is in order.

The database was originally built years ago in Progress (old OpenEdge) but some Clients run it through the OpenEdge SQL DataServer which basically migrates the schema and data into SQL and then uses a schema placeholder to translate OpenEdge database queries into SQL queries. This means that the database can run on SQL whilst the main GUI and interface doesn't need re-coding and can continue to run in Progress 4GL.

There genuinely are no (clustered) primary keys and very few clustered keys; most indices are non-clustered indices that yes will just exist on the heap. It's not a case of them trying to keep the db a secret as it is fairly self-explanatory; I used to work for the supplier and know the db inside out.

I'd like an idea for what kind of impact having no primary, clustered keys at all would have on your typical transactional database...