How are you sharding the TPC-C workload?
If the answer is sharding on warehouse, how are you implementing the distributed transactions in new order?