We recently had some trouble with our production SQL 2008 R2 databases and had to failover to log-shipped warm spares for most of our databases. This morning I found some very odd GUID values scattered throughout our set of databases. For background, we have a main clients database, that holds various information about installed clients, including the master client_guid that is referenced in other tables and other databases. I am seeing that the GUIDs in some of the support dbs have become corrupted with Kanji and Cyrillic characters. For example: In our main client data table, a particular record had a GUID of '4d86854e-d699-4bce-a98b-c34fcc909453', but in the Analytics database that same GUID is showing up as '4d86854e-d699-4bce瞧ＲĹ（Ɏ-c34fcc909453'.
I have been wracking my brain today trying to figure out how this might have happened. I stumbled upon some information about Unicode compression, and I was wondering if it might be possible for SQL Server to mangle these GUIDs during a database restore.
Collation throughout our db system is set to SQL_Latin1_General_CP1_CI_AS.
I'm wondering if anyone has any insight into this problem.