Re: IBM Informix IDS Benchmarking.
Neil Truby argued IDS lacks TPC benchmark results because repeating costly benchmarks is prohibitive and IDS is not IBM's flagship database. Mark Townsend replied the main cost is hardware and storage, borne by the hardware vendor, here IBM. Another poster said hardware is a small part of the cost; Townsend answered disk storage dominates, 40-100 TB for TPC-C. Unresolved.
Auto-generated by Claude from the posts below — may be imperfect; read the full thread.
"Fozzie The Bear" <sorry@no.com> wrote in message news:1YTEa.237$FL1.4096@news-server.bigpond.net.au... > There seems to be a push on for DB2 benchmarks from recent review of tpc.org > although SQL Server seems to just beat the posted mark 10 days later. > The Qn. is why don't IBM Benchmark IDS / XPS against ee / 8.1 ... Cost. Once you've spent the cash on one of these benchmarks, then your product - say IDS - is top. But then, over the coming months as hardware etc imrpoves, it starts to slip down, until 12 months later it's, say, 10th. This is a baaaad situation - much worse than if you'd never participated at all, and therefore had no ranking. And so, you have to spend the (millions of) dollars again to re-benchmark. IDS is not IBM's strategic, flagship database. Therefore cost prohibits it being submitted for benchmarking.
Neil Truby wrote: <snip> > IDS is not IBM's strategic, flagship database. Therefore cost prohibits it > being submitted for benchmarking. > Umm - the cost is in the hardware and storage. So it's typically the hardware vendor that bears the main cost in the benchmark - and in this case, that is of course, err, IBM ? Isn't this like Goodyear saying they won't run a race car at the Indy 500 because they can't afford the tires ?
Mark Townsend <markbtownsend@attbi.com> wrote in message news:<HveFa.1192050$S_4.1213179@rwcrnsc53>... > Neil Truby wrote: > > <snip> > > > IDS is not IBM's strategic, flagship database. Therefore cost prohibits it > > being submitted for benchmarking. > > > > Umm - the cost is in the hardware and storage. So it's typically the > hardware vendor that bears the main cost in the benchmark - and in this > case, that is of course, err, IBM ? Isn't this like Goodyear saying they > won't run a race car at the Indy 500 because they can't afford the tires ? So you put some boxes with the new hardware in in the room containing the box and the software installs itself, runs the benchmark and configures itself? The cost of the hardware (especially PC-style hardware) is only a small part of the benchmark costs.
5 cats wrote: > The cost of the hardware (especially PC-style > hardware) is only a small part of the benchmark costs. The biggest cost is and has always been the disk storage required. Note that a top end TPC-C will use between 40-100 Tb of storage, no matter what type of CPU used.