Re: FW: FREE ORACLE SOFTWARE!

ptash@iconz.co.nz (Peter Tashkoff) wrote: >In article <3tbj75$s6d@inet-nntp-gw-1.us.oracle.com>, tkyte@us.oracle.com >says... ..... >>Where is the cheap shot here. PC Mag. designed it (using a modified >>AS3AP standard benchmark), ran it, reported it... Had nothing to do >>with any one database vendor. If Informix had any complaints they >>could have just withheld the results by never doing to test at all, >>the manner of testing and platforms was known well in advance. They >>took a chance and lost. >> >>Thomas Kyte >>tkyte@us.oracle.com >> >Took a chance and *lost*!? >Normally I am not a database partisan but am I reading this correctly? >Is this fellow really claiming some >kind of kudos for Oracle/Unix outperforming Informix/Dos? >Well hey, claim your small victories guy. If that's all you've got. >Peter Tashkoff >Auckland, NZ Well lets see. You think that OF COURSE Oracle on Unix should blow away Informix on Dos. Lets look at the *facts*: 1993 - PC magazine runs evaluation. PC magazine picks hardware. Database Vendors Pick OS. Lets see what was choosen in 93: Oracle - Novell - v7 Informix - Novell - v4 Ok, same hardware, same OS, no problem. 1993 - Same constraints. PC Mag picks hardware, Database Vendors picks OS. Lets see what was choosen again: Oracle - Novell - v7.0.16 Informix - SCO (that would be unix wouldn't it?) - v5.01 So actually this is some kind of kudos for Oracle/NOS outperforming Informix/UNIX. Note that DOS (MS or PC) was never used by anyone in any of these tests. Windows-NT, Novell, SCO, and OS/2 were the operating systems used in the 1993-94 timeframe. So I guess if you think that of course a database running on UNIX will outperform a database running on 'DOS' (which perhaps you confused Novell NOS with DOS), then this is pretty good... And yes, 'Took a chance and lost'. They opted to participate. They supplied the software (but no people to back it up, PC-Mag hired an Informix consultant from Open Systems Engineering to perform the tests). Since no one can publish benchmark results for the databases without the expressed written consent of the database vendors (makes sense, you wouldn't want Sybase to contract to a third party to benchmark themselves against Informix and then publish the results as an independent benchmark would you?) Informix took the chance, and they lost. If they didn't like the test in the first place, they didn't have to participate. Informix picked the OS they wanted on the given hardware. The hardware was the same, we all had the same RAM, controllers, Drives, network cables, etc. The only difference between any of the database vendors execution of the test were: 1.) the operating system choosen 2.) the database software used Everything else was held constant. The database vendors chose the above. Would Informix pick the OS they would run slowest on given the hardware? If they did, that was their mistake I guess. Thomas Kyte tkyte@us.oracle.com Oracle Government