fr den 03.03.2006 klokka 07:40 (-0500) skreiv Nicolas Ross: > Hi ! > > Currently, I have a linux router with a gigabit trunk to a 10/100 swith. > This trunk passes vlan tag to different vlan interface in the linux box. > > We will be adding a 24 port gigabit swith soon. We then will have a gigabit > segment. On that segment, we plan on unsing jumbo frames (9000 or the like). > So, I'll have to crank up the eth undelaying card of the linux router to > 9000 to support such packets. New vlan interface will default then to a mtu > of 9000 (I think so). In practice I don't think this will give you a lot performance-wise. The most of the world is at 1500 bytes so the MTU in use at most of the time would probably be 1500. This could work in a scenario where you are doing large transaction where window scaling max out to 9000 bytes. > But we have some other vlan that needs to stay at 1500. If I specify in the > ifconfig command the MTU, will this work as I expected ? yes > Also, is there a way to specify de MTU in the ifcfg-vlanxxx file in > sysconfig/network-scripts ? at least in Fedora, yes :)