Ubuntu Server with LTSP in real life

[Bart Silverstrim]

On Mar 30, 2007, at 9:05 PM, Michael T. Richter wrote:

> On Fri, 2007-30-03 at 03:37 -0500, Brian Fahrlander wrote:
>> Second, Paul, that's very cool; I started in LTSP back when  
>> Etherboot had just changed over, probably about 1996-97. I had 12  
>> users running old tired 486's in a lumber yard, and only a couple  
>> of things caused me grief:
>
> Aren't diskless workstations just another word for "single point of  
> massive failure"?  I've never understood the appeal in this day and  
> age where you can get old P/P2 (and even P3) hardware complete with  
> disk for a song and a dance.  What am I missing?

Management, easy deployment of software, and when we ran terminals,  
an easy way to remotely troubleshoot and also replace old systems  
without losing people's preferences and personalizations.

If you set up a central server with RAID and good backups, or even  
use two machines in cluster with heartbeat, you can minimize downtime  
as a single point of failure.  I mean, in most setups that are  
managed, if your server dies your workers are still up a creek  
without a paddle until you get the server back up.  With terminals it  
just means they can't play solitaire while waiting for you to get the  
server back up.

That's how it was with what we once ran.  It also meant that if you  
have a high end server and your workers are in areas of high dirt or  
other not-so-computer-friendly areas, they could run applications at  
high speed (since it was running on a high-end system located  
elsewhere) while the cost of the hardware at their location is table  
scraps in the budget.  Even a P/P2 or P3 with a disk you picked up  
"for a song" is hard to compete with that, once you factor in  
installing software for field deployment when the hardware starts  
failing partially due to conditions.

Just all depends on your environment and needs.

-Bart