Teradata's architecture uses AMPs (Access Module Processors) that each own a portion of data on their associated vdisks - they access this data independently when queries require it. The PEs (Parsing Engines) handle query parsing and send requests to AMPs through the BYNET, which is the high-speed interconnect that enables communication between components. Option A describes disk areas but isn't an architectural attribute, while option B is incorrect because Teradata uses hardware parallelism, not OS time-slicing.