The standards document says that 64 or 68 bytes is acceptable for a minimum tagged packet size, and that if a bridge untags a 64 byte packet it must pad it out. I wonder what the performance implications would be for a bridge to do this, and if it would be better from an overall network performance point of view if the linux vlan implementation did insist on a 68 byte packet length when transmitting... If it's all in hardware it shouldn't matter. James > -----Original Message----- > From: vlan-bounces@xxxxxxxxxxxxxxx [mailto:vlan-bounces@xxxxxxxxxxxxxxx] > On Behalf Of Ben Greear > Sent: Thursday, 31 August 2006 12:27 > To: Linux 802.1Q VLAN > Subject: Re: [VLAN] 802.1Q ARP frame size? > > James Harper wrote: > >> John C. Lin wrote: > >>> (Sorry, if this is a FAQ) > >>> > >>> I noticed that the 802.1Q ARP frame size generated by the Linux > > driver > >>> (V1.8) is only 64 octets (not 68 octets). > >>> > >>> It seems the minimum frame size for 802.1Q is 68 octets. > >> The 64-octets is an ethernet physical level issue, not a VLAN issue, > > so > >> there is no need to pad the frame to 68 bytes. > >> > > > > So what happens when a switch untags the vlan tag and sends the untagged > > packet out a port? Is it the responsibility of the switch to then pad > > the packet? Or does this just happen automatically anyway? > > It's the switches duty..and most NIC's hardware will do the padding..I > assume switch hardware can easily do the same. > > Ben > > > > > James > > > > _______________________________________________ > > Vlan mailing list > > Vlan@xxxxxxxxxxxxxxx > > http://www.candelatech.com/mailman/listinfo/vlan > > > > > -- > Ben Greear <greearb@xxxxxxxxxxxxxxx> > Candela Technologies Inc http://www.candelatech.com > > _______________________________________________ > Vlan mailing list > Vlan@xxxxxxxxxxxxxxx > http://www.candelatech.com/mailman/listinfo/vlan
