El mié, 12-08-2009 a las 13:31 +0200, RalfGesellensetter escribió: > Am Mittwoch 12 August 2009 schrieb nigel barker: > > Jose suggested the NIC is the bottleneck > > Hi Nigel, I'd support this suspect. Even sticking to the same hardware > and (more or less) software, I am experiencing 100% netload on eth1 > (serving 100 Mbit). > > I ascribe this to > - new TFT screens with higher resolution (more X data to serve) > - increasing use of Flash videos > - increasing use of animations all over the web > - possibly also more use of Java related stuff (is this rendered > by native X11 sequences or rather as direct graphic drawing?) > > Whenever a Mbit NIC or a switch with Mbit Uplink is not affordable, > I wonder if this could be an option (howto required): > > Give your LTSP yet another 100Mbit NIC and bridge eth1 and eth2 to serve > your thin clients by 200 Mbit! (how exactly?). > That's called "bonding". Take a look at http://www.debian-administration.org/articles/350 Anyway, 200 Mbps is not enough for 20 clients (well, it might work but they will boot slowly and won't be able to reproduce some Impress animations, and probably, web sites). Just as a reminder (with my not-very-scientific-or-accurate-measurements in Lenny), the server must be able to provide: 34 Mbps x No of LTSP clients Cheers José L.
Attachment:
signature.asc
Description: Esta parte del mensaje =?ISO-8859-1?Q?est=E1?= firmada digitalmente