Re: Beginner help needed in database design

"DA Morgan" <damorgan_at_x.washington.edu> wrote in message news:1107027368.409837_at_yasure...
>
> Anytime you find a benchmark that compares apples with apples feel
> free to post it. Until then please stop whining. You do recall that
> it is you that started this.
> --
> Daniel A. Morgan

That would be nice (apples to apples). But Oracle RAC (using multiple clustered nodes with 64 total processors) is a different architecture what DB2 uses in the TPC-C benchmark (using a single SMP node with 64 processors), so that is not possible. Of course, Oracle clearly markets RAC (and legitimately so) as something that DB2 does not have (even though it is comparing apples to oranges).

Oracle markets RAC as an architecture that scales well with lots of cheap nodes in an OLTP environment. Since the TPC benchmarks measure both total TPM (transactions per minute) and cost/TPM, Oracle could presumably just increase the number of cheap nodes to account for the fact that they processors are not the fastest available. If Oracle did use the fasted IBM eServer p5 595 in a multi-node environment (say 16 nodes with 4 processors each), the cost/TPM would be significantly higher than the IBM/DB2 results on a single 64 CPU node.

In any case, Oracle and Mark Townsend certainly have advertised Oracle benchmarks wins on hardware significantly different than used for the latest equivalent DB2 benchmarks, without any mention of the hardware difference (or that the DB2 benchmark is 2.5 years old). I think it is up to reader (without the hyperbole and viciousness display by Daniel Morgan on this newsgroup) to take these things into account from both vendors. But it will always be an apples to oranges comparison, so customers will just have to figure it out. Received on Sat Jan 29 2005 - 14:22:35 CST