On 3/9/2011 9:55 AM, Brunner, Brian T. wrote:
This is where mental ossification amongst bean-counters can kill a company. "Economic Opportunity Cost" should raise its head here: What would we do with the $capex if we paid $opex vs what would we do with the $opex if we paid $capex. "The Time Value of Money vs The Money Value of Time" is another phrasing of this point-of-view. Unfortunately this is no longer a CentOS topic.
The admin/operator's time is usually seen as a fixed cost and keeping a machine working is not supposed to take unplanned time. So, if you want to keep something running you really need to buy 3 of them in the first place. One as primary in production, one as a backup, and one to be developing/testing the next version on. In some cases you can replace the third one with a virtual setup, and you might be able to have one backup as a spare for more than one live server but you can't skimp much more than that. Everything breaks, so if one thing breaking causes a big problem, it wasn't planned realistically. This should be a 'swap in the backup' while you run extensive diagnostics or get a warranty repair on the broken thing. And if you are running Centos the one thing you don't need is to pay for extra licenses to cover the backup/development instances.