Has anyone gone down the path of trying to set up partitioning on CRM tables for performance/archive reasons in very large CRM DB's? Would this even be supported? To explain, say you added a field called 'archive' to opportunities, then set up table partitions such that when archive = 1, it would go to one SQL partition, and when it is 0/null it would go to the 'other' SQL partition.
Then you change your views to all include the 'archive' attribute as filter criteria (separate views for 'current', 'archive' , and 'all' data as required) - at that point your 'non-archive' views are querying against a potentially vastly smaller 'active' dataset, which could also be targeted to be stored on separate/faster disks as well.