How best to store multi-dimensional timeseries data in a (R)DBMS environment?

Any suggestions on how multi-dimensional time-series data should be stored and organized in an RDBMS environment? For example, we have multiple experiments from multiple customers. This translates well to say a NetCDF file type structure, where one NetCDF can be for one customer project, and it contains several data variables overtime. This image explains the concept. From my preliminary research , this seems like an unsolved problem or one with a complicated solution . Are there any recent (not necessarily optimal) attempts at this?