> > I was thinking about the approach of separating > > servers for different applications, meanning that if > > the server that serves the office app dies, it will be > > the only app not available until the server reboots, > > without killing the complete sessions. > > Not necessarilly a good idea. Of course, your X server can display > multiple applications from multiple simultaneous servers. That's a major > design goal of X. However, it's much better to provide failover for all > applications by clustering them for failover. That way, if one server > goes down, the others keep the application available.
I agree. It's also a *lot* easier to maintain a single disk image and clone it to new servers as they are needed. The load balancing and fail-over system I've developed supports a primary and secondary where the two system images differ only in the contents of dhcpd.conf. Garry Saddington recently described a clustering system he's been using where the servers are identical. Both of our approaches support the idea of a "master image" that can be used to quickly repair a failed application server or put a new one on the network with a minimum amount of configuration. We also don't store user data or state on the LTSP servers... which means we don't need to back up or synchronize them. On the systems I've put in, all user data is saved to a mysterious file server that someone else is responsible for backing up! :) -Tom ------------------------------------------------------- This sf.net email is sponsored by: See the NEW Palm Tungsten T handheld. Power & Color in a compact size! http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0001en _____________________________________________________________________ Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto: https://lists.sourceforge.net/lists/listinfo/ltsp-discuss For additional LTSP help, try #ltsp channel on irc.openprojects.net
