Forum: Other

Post: TPC Benchmarks?

TPC Benchmarks?
chbussler
Apr 13, 2010
Hi,

when I read the academic papers about HStore, then there is reference made to the TPC benchmarks. Do you plan to run those and publish the performance report officially with http://www.tpc.org/?

Thanks,
Christoph
TPC Benchmarks?
bheath
Apr 13, 2010
Christoph,

Thanks for this thought, we are indeed proud of our performance results. However, what we are doing is similar, but not identical to the strict implementation of the tpc-c benchmark. That's why we refer to it as "tpc-c like". The tpc-c benchmark has some requirements, such as wait time, that we feel are no longer relevant to today's OLTP environment. We have no plans currently to submit results to the Transaction Processing Council.

- Bobbi
'TPC-C' like results?
chbussler
Apr 13, 2010
Christoph,

Thanks for this thought, we are indeed proud of our performance results. However, what we are doing is similar, but not identical to the strict implementation of the tpc-c benchmark. That's why we refer to it as "tpc-c like". The tpc-c benchmark has some requirements, such as wait time, that we feel are no longer relevant to today's OLTP environment. We have no plans currently to submit results to the Transaction Processing Council.

- Bobbi


Hi,

thanks a lot; I can see people asking for a comparison; I am sure you get this question a lot:-)

Not sure if it is possible to characterize the difference between the 'real' tpc-c and the 'tpc-like' benchmark and provide results in the sense that you described by listing what does not apply any more. But from the angle of running the required amount of client requests, one certainly could compare (i.e., time/cost to arrive to the same functional result, even though the implementation is not comparable).

Just thoughts.

Thanks,
Christoph
Seconding
henning
Apr 16, 2010
Hi,

thanks a lot; I can see people asking for a comparison; I am sure you get this question a lot:-)

Not sure if it is possible to characterize the difference between the 'real' tpc-c and the 'tpc-like' benchmark and provide results in the sense that you described by listing what does not apply any more. But from the angle of running the required amount of client requests, one certainly could compare (i.e., time/cost to arrive to the same functional result, even though the implementation is not comparable).

Just thoughts.

Thanks,
Christoph


It would be cool to learn where the 50x claim originates from, in source.

Regards,
Henning