| TPC-D design contains flaws that make its use questionable even for sequential database systems |
| Given the complexity of the parallel database systems, attempts to reduce their performance estimate to one number are most likely futile |
| TPC-D defines no metrics useful for understanding behavior of parallel RDBMSes. TPC-D designers should re-examine academic research work pertinent to the field |
| Future research work towards definition of DS benchmark for parallel databases seems important |
