the weekly whinge

This Dr Dobbs article caught my eye when I was looking for a Windows Mobile NTP client (long story). I read it with some alarm as the introduction implies that they’ve found a fundamental problem with NTP and then became steadily more incredulous as I read the article to find that they were talking about periodically running – on mission-critical systems, of course – ntpclient to synchronise with a single time source.

They’ve basically written up alarmist crap that’s worthy of the Daily Mail about the fact that if you use what’s essentially a test program against a single not-particularly-stable server it doesn’t work very well. Big surprise.

Does anyone really use ntpclient like that on a critical production server? I’d have thought configuring ntpd with a few ntp servers wasn’t exactly hard – there’s even a pretty noddy Redhat howto for it. And yet the title of the article (and especially the introduction) suggest that they’ve uncovered a fundamental problem with ntp.

I’m left wondering if timekeeper is a Dr Dobbs product, to be honest.