Sql-server – What’s the best solution for distribution of one database over two servers

64k blocks are usually best. Log writes should actually max out at about 60k, but at 64k you are aligned with the size of the physical block on the disk.

If you're only concerned about Agent Jobs and logins, I think the easiest approach is to leverage a linked server and querying the relative tables. I would do this by first creating a linked server on the old instance that points at the new instance. Once I had that, I could run the following queries:

Logins

SELECT * FROM [SQL2005].[master].[sys].[server_principals]
WHERE name NOT IN (SELECT name FROM [SQL2008].[master].[sys].[server_principals])

Agent Jobs

SELECT * FROM [SQL2005].[msdb].[dbo].[sysjobs]
WHERE name NOT IN (SELECT name FROM [SQL2008].[msdb].[dbo].[sysjobs])

There might be some inconsistencies with some server specific accounts/logins, but these queries should show you everything that exists in the old server that doesn't exist in the new server. You should be able to extend this logic to other facets as long as you had system tables to query for it.

As for objects that need to be synchronized, I think the best approach is to review your instance for the objects that should be migrated. Typically this would only be logins and Agent Jobs, but could also include linked servers (sys.servers), end points (sys.endpoints), and other objects.