Re: TPC-C implementation

Volker Hetzer wrote:
>
> "Kai Honsel" <kaiho_at_uni-muenster.de> schrieb im Newsbeitrag news:bob41t$16dk$1_at_redenix.uni-muenster.de...
> > Hi everybody!
> >
> > I'm searching for an TPC-C implementation that I want to (re-)use for
> > educational purposes. Does anybody know where I can get it? (The TPC does
> > not provide a complete implementation of TPC-C.)

Are you really sure you want to go down that path?

Rumor has it that: a) vendors don't allow discussion of non-vendor-controlled benchmarks (and your 'education' could be misconstrued as a benchmark unless you control access very carefully); b) vendors typically spend $millions on tuning each TPC result; and c) every implementation is unique!

I looked down that path about 5 years ago. My conclusions: what I really wanted was a database big enough to do some serious tuning (see asktom for some ideas on making a big data set); TPC is really not very meaningful unless I wanted ficticious/idealogical CPU to $ relationships; the general cost of establishing the baseline environment for any meaningful run is enormous; I would end up violating every license agreement I had at the time.

Aside from that, my understanding is that the TCP details are IP of TCP.org and must be appropriately licensed.

However, your conclusions may vary because your situation is different than mine and/or because other things may have changed.

All that said - check out the appendix sections in the full disclosure documents. (You might particularly enjoy wading through the 'create' scripts and the 100's of 'create rollback segment' definitions.)

> What is TPC?

http://www.tpc.org/

>
> Lots of Greetings!
> Volker

-- 
/Hans
[mailto:`echo $from" | sed "s/yahoo/telusplanet/g"`]