We're in the process of planning the migration of some databases from an old doddery server (running SQL Server 2000) to a new one running SQL Server 2008. As not everything is going to the same new server, we're intending to 'retire' the name of the old server and use a new one more in keeping with our current naming conventions. But, due to various reasons of bad practice and auditing, I don't believe we'll be able to definitively identify pre-migration absolutely everything that may still refer to the old server name. This means that for a while it's likely we'd have to redirect any reference to the old name to the new name and hope nothing breaks (most migrated DBs will be going to this server, so we'll likely play the odds).
With this in mind, I was wondering what would be the best way to detect any connection attempts that have used the old name? Is this something that can be done via SQL Server, or only via the internal DNS servers? The former is preferable to me as I can monitor this myself. Would the latter be able to detect the exact connection string used (which would help identify the application using it)? I think it's likely that many of the redirected connections will involve either Crystal Reports or Access ADPs.
A possible complication is that our network is Novell, not MS, although the new SQL Server will be on Windows Server 2008.
Any advice is welcome.
I'm not a pro at using Ethereal, but your best bet is probably using a packet sniffer and filtering on IPX.