[ntp:questions] Re: Mother Of All Clocks

Heiko Gerstung heiko.gerstung at meinberg.de
Tue Mar 29 05:59:13 UTC 2005


Todd Knarr wrote:
> In comp.protocols.time.ntp <88e50c98.0503282017.1082ad81 at posting.google.com> Heiko Gerstung <hg at heiko-gerstung.de> wrote:
> 
>>[...]
>>on your side and these guys can get really unfriendly if you are
>>trying to touch a cable...
> 
> 
> But what about the guy who's _supposed_ to be touching the cables, since
> that was what he was called in to do?
> 

We are offering this kind of service, too (touching cables ;-) ...) and 
we are held reliable for anything that happens while we are in that 
server room.

> I've rarely run into an outage caused by unauthorized people or just
> plain accidents. If they're not acts of God (or backhoe operators),
> they're almost always caused by people who were completely authorized to
> be in that equipment working on it, were in it on a job they were
> supposed to be doing, and were absolutely sure they were doing exactly
> the right thing right up to the point where something critical went
> dark. Physical dispersement at least gives you the time it takes for him
> to walk from one rack to another to catch the alarms and stop the
> failure before more than a couple or three networks are hosed. I factor
> in not only the chance of a failure but how catastrophic the failure
> will be if (or more likely, when) it happens.

Of course this happens, but if someone has to work on something in a 
sensitive network, he/she _should_ be aware of the danger and _should_ 
have prepared the rest of the world to be on service at least while the 
work is in progress (please note the emphasized "should"'s)

So if someone shows the guy how many years he needs to work to cover a 
possible damage, this very often helps to ensure that he pays attention.

In this specific case a second monster is there to take over if things 
really go badly wrong.

> 
> As far as being paranoid, sysadmin's rule: "It's not whether you're
> paranoid. It's whether you're paranoid enough."
> 

Your point :-)

Kind regards,
Heiko



More information about the questions mailing list