On Apr 26, 2017, at 8:47, Cowboy <[email protected]> wrote:
> Hey, I know !
> Let's take something that works and has worked for decades perfectly fine,
> ( init ) and whole-sale replace it, because we can !

Without speaking for Rob, and not knowing his specific situation, my guess is 
that it's called preventative maintenance: replace the system hardware before 
it fails.  Those spinning disks don't last forever, after all, and neither do 
power supplies.

This issue is one where probability and statistics mean that the farther out 
you go on the distribution curve (i.e. the longer in time you wait), the higher 
the cumulative probability that you'll suffer a system failure.

The challenge for the broadcast engineer is knowing when to replace a 
seemingly-perfectly-functioning system with another brand-new one that has no 
known history (other than the initial burn-in).  And, of course as they tell 
you in financial markets, past performance is no guarantee of future 
performance -- so that new system could fail in a day, or the old system could 
last beyond your time at the station (or on this earth, but that's another 
story).  

But generally, it's a risky thing to keep 10-year-old computers running 24x7 - 
primarily if they were built with consumer-grade hardware.

Yes, we have 10-y-o servers running at my workplace, and I've only had one 
drive fail out of 48 SATA disks in a couple of NAS units around that vintage, 
but stats are stats ... 


The real question here is whether we would rather deal with hard downtime after 
a unit fails, or proactively swap it out prior to full systemic failure.  

The other option, of course, is to put the new system in place in parallel with 
the old and have a hot standby system.  That's the way we ran our Rivendell 
system for ~8 years ... but doing that does mean that maintenance doubles, and 
workload (i.e. normal operations usage of loading up new content, schedules, 
etc.) also doubles.

To each his own style and design... as to whatever makes the most sense in 
their environment.
—
Sherrod Munday
<[email protected]>

_______________________________________________
Rivendell-dev mailing list
[email protected]
http://caspian.paravelsystems.com/mailman/listinfo/rivendell-dev

Reply via email to