So you've put about 200 man hours into your configuration system this year. It's good. There are a number of things you'd change about it, but you're at a point that it's running rather smoothly. All of a sudden, there's this great new system that comes out. All of your problems are solved! It promises to slice, dice, and give you an infinite amount of differing system class configurations, with a new web 2.0 way to group systems.

DONT DO IT! YOU HAVE TOO MUCH TO LIVE FOR!

Seriously here, it's just hard to learn how to say "this is done" and move on. For me, it lays in these four areas:

- ISP
- Monitoring
- Configuration Management
- Hardware

There's always a better piece of hardware out there, even though your existing systems meet your needs just fine. There's always an ISP with a slightly cheaper rate, but you're not doing BGP so you'd have to re-ip. There's a great new monitoring system with an AJAX interface, but you've put 10% of 2006 so far into making your Nagios/Cacti solution know when to page, and it's finally all green!

It feels that in our line of work, nothing is ever done. It can always be tweaked, made better. Quite a few of our projects start out with terribly vague goals, no set milestones, and no defined ending. Somebody says "We need a monitoring system".. So you start monitoring. Every week or so, you find more things you can shove into your monitoring system.. Trend Analysis, performance graphing, clustering, SMS paging, AIM paging, flashing stoplights.

Ack, there's the door.