I have a client that would like to use "tagging" to identify documents that are related to specific customer/case data. Some terms would be customer identifiers with child terms for customer cases. There are more than 1K customers so each customer cannot be its own term set and there are ~500K such cases. It seems that within the managed metadata store limits (7 nested levels, 1,000 term sets, 30,000 terms per set, 1M max number of terms) we could come up with a hierarchy for these entities as terms.
Is this a valid way to structure managed metadata?
What are the repercussions of trying to use like terms across term sets?