A lot of the fun of the MCM lab was that the questions had surprising catches. The questions and answers alone weren't enough - the lab environment itself had tricky gotchas that you had to solve in order to get to the solution.

This wasn't one of the questions, but it has the same spirit. Say I gave you two virtual machines, each of which were running their own SQL Server instances, with transactional replication flowing between them. Your question would involve adding an index to one of the databases, but you'd have to know that doing it would break replication, and you'd have to know how to work around it. (Especially if somebody had already made hand-coded changes to the replication stored procs.)