El jue, 17-04-2008 a las 20:14 +0200, RalfGesellensetter escribió: > Am Donnerstag 17 April 2008 schrieb José L. Redrejo Rodríguez: > > We have discarded that use, we want to have low power comsumption and > > 0-maintenance thin clients (think we're dealing with several > > thousands of them, so those characteristics are really important). > > Hi José, misunderstanding? > sure ;-) > I meant: to boot the thin client SERVER via PXE from a super-LTSP > Server, kind of cascading LTSP servers ;) Not,I didn't tested it. It's an attractive idea from the maintance point of view, but that would force us to change part of the current infraestructure in the schools. The communication between that super server and the rest of the servers in the school should be via a 1 Gbps backbone, it would mean changing a lot of switches and cables in the schools (unless I misunderstood you again). I can not see any perfomance improvement using this topology. We're using puppet in our tests to assure self-configuration and updates of all the ltsp servers in the school, and it works pretty well. Obviously, making a config change does not propagate as fast as using the super-ltsp server. Cheers.
Attachment:
signature.asc
Description: Esta parte del mensaje =?ISO-8859-1?Q?est=E1?= firmada digitalmente