> > What kind of ninny takes a single number from a lab-queen test and uses it to build > > the pricing model for their business?
> At least two of my previous employers. Of course one is no longer a > going concern, and the other is presently on the ropes. Of course > neither of them have ever sold Linux or S/390 software, so I guess that > makes 3 ninnies. I spoke with (I think) Ed Gauthier about this at an analysts' meeting in April or so. I think the attitude was pretty much "You can do some dumb things with a machine if you don't need it to do any work, and we can get pretty bored in coffee breaks". There was a similar exercise a few years back when someone tried to run n levels of VM under each other, but apparently SIE made that pretty pointless. They're stunts, justifiable because every now and then something will jump out and eat your lunch - so you can fix it before scalability really does reach those levels. And it will, one day. I just got a pre-release of some IBM DB2 UDB pricing for Linux - the ratios between (large) Windows or other Linux servers and 'Enterprise' servers seem pretty sensible - fingers of one hand sort of numbers. Those who try to bring their old practices into the Linux world deserve to go to the wall. We can wave as they go past. -- Phil Payne http://www.isham-research.com +44 7785 302 803
