Re: Oracle TPC Benchmarks and "discrete transactions"

In article <wwknp=p.tcox_at_netcom.com>, tcox_at_netcom.com (Thomas Cox) writes...
>devine_at_olympus.CS.Berkeley.EDU (bob devine) writes:
>>It is obvious that Oracle has invented discrete transactions as a
>>way to generate higher TPC numbers.
>
>Actually, I was under the impression that Discrete Transactions are
>found in IMS/Fastpath. Can someone confirm this?
>

From the description of the limitations of discrete transactions in ORACLE documentation, I conclude they are not nearly as useful as the IMS/Fastpath feature. In particular, since you can not perform SELECT FOR UPDATE in an ORACLE discrete transaction it is not possible to express "If QTY-ON-HAND > 0 THEN QTY-ON-HAND = QTY-ON-HAND - 1" and have it actually work. With normal SELECT statements, ORACLE will give you the value of QTY-ON-HAND at some point in the past and will not lock it against changes by other transactions.

>>While these probably don't
>>break the letter of the TPC rules, they do break its spirit!
>
>If dt's are useful in a TPC environment, then they will be useful in
>those real world applications that resemble the TPC environment. If
>there are no such applications, then the TPC is useless. Which position
>are you taking?

If "resemble" were true there would be little argument. From Charles' posting and my own reading of the Oracle material I conclude that the environment must be almost identical for them to be useful. That's one of the things Charles is asking people...could you use discrete transactions in your application. Further, if you can almost use them but not quite (do to a particular restriction) then I'd like to hear about slight variations that would be lots more useful.

As for the usefulness of the TPC-A and TPC-B benchmarks...they had a purpose at one point in time. It may very well be that they have outlived that usefulness for a number of reasons. This is not new in the benchmark world. At various points in time USSTEEL, WHETSTONES, DHRYSTONES, and other relatively limited scope benchmarks were "the" measurements of performance. They have all faded from the radar screen, or at least become parts of more comprehensive measurement methodologies. Even some of those recent methodologies, SPEC89 for example, have had to be replaced (SPEC92) to maintain a reasonable basis for comparison.

.............................................................................

Hal Berenson

Home: 71640.3535_at_compuserve.com
Work: berenson_at_cookie.enet.dec.com

• Disclaimer: Opinions expressed here are my own, not my employer's! If I happen to communicate with you from work rather than home, its just for convenience (just like asking for a "daytime phone number") and should not be construed as representing the views of my employer or its employees, officers, directors, or stockholders. --