folks, keep one thing in mind in the process of the PC packets getting to the switch infrastructure. the IP phone is a switch itself, so as packets come in port 0 (PC) and go out port 1 (to switch) those layer 2 headers are 'touched'. So, as it gets forwarded upstream (onto a 802.1q trunk, which is the only place that COS can exist) those packets can be marked (rewritten) with the appropriate COS value.
On Wed, Jul 23, 2008 at 10:01 PM, Nick Marus <[EMAIL PROTECTED]> wrote: > COS does exist in the 1q header in the 1p field, however the pc nic > (most) has the ability to add this to the frame. (should see "qos packet > scheduler" under you nic properties in windows) The 1q header can exist with > out vlan tagging being set. unlike isl which re-encapsulates the ethernet > frame inside an isl frame, adding the 1q header does not render the frame > unreadable to non 1q devices. The 1q header is just inserted between the sa > and the type/length field in the frame header. many devices that do not make > use of 1q vlan tagging still use the 1p portion of the 1q header to mark > their cos. > > A cisco switch that is setup for mode access will ignore the 1q vlan > tagging and put the frame on the vlan that the switchport is configured > to. But it will still look at the 1p portion of the 1q header if qos trust > cos is configured on the port. Same thing goes for a computer attached to a > phone. if the switch is configured to trust the cos marking of the device > attached behind the phone, it will look at the 1p portion of the 1q header > on the frames coming from the device attached to the phone and respect those > values. otherwise it will have the phone remark those values before the > frame gets to the switch and get prioritized. > > <http://www.cisco.com/en/US/tech/tk389/tk689/technologies_tech_note09186a0080094665.shtml#topic1> > > > > > > On Wed, Jul 23, 2008 at 6:57 PM, Mike Brooks <[EMAIL PROTECTED]> wrote: > >> Its my understanding that the CoS value is set in the 802.1p field >> within an 802.1q tag. Therefore, in order to set a CoS value you need >> have an 802.1q trunk. So a PC would not be able to set a CoS value, >> unless its uplink was an 802.1q trunk port, rather than an access >> port. >> >> So if the PC can't set the CoS value, why would you need to use the >> "switchport priority extend cos 0" ? >> >> Please correct me if I am wrong. >> >> Regards, >> >> Mike Brooks >> CCIE#16027 (R&S) >> >> >> On 7/23/08, Nick Marus <[EMAIL PROTECTED]> wrote: >> > Unless I misunderstand you, COS is applied to a packet and does not >> require >> > a 1q trunk. The connection between the pc and the phone is not 1q >> usually. >> > Just the connection between the phone and the switch. Most PC nic's can >> be >> > setup to mark it's packets with a cos value and effectively take >> priority on >> > your switched network over you voice and other high priority packets if >> the >> > switch is trusting and the phone is not remarking to 0. >> > >> > >> > Nick >> > >> > >> > On Wed, Jul 23, 2008 at 10:51 AM, Mike Brooks <[EMAIL PROTECTED]> >> wrote: >> > > Hi Everyone, >> > > >> > > I see that the standard practice on a switchport is to configure >> > > "switchport priority extend cos 0" in order to allow the ip phone to >> > > reset the cos value received from the PC to 0. >> > > >> > > My question is how would a PC ever set a "CoS" value if the link >> > > between the ip phone and the PC is not an 802.1q trunk ? >> > > >> > > Can someone please help me understand this ? The only thing I can >> > > think of is that the PC would somehow have to support an 802.1q trunk >> > > to it, a trunk would have to be dynamically established between the >> > > phone and PC. And, then the user would have to manipulate the CoS >> > > value. Is this possible with a Cisco phone ? >> > > >> > > If this is the only case this would work then you would think that >> > > Cisco would document these pre-requisites. Perhaps I am confused. >> > > >> > > Please help ;-) >> > > >> > > Regards, >> > > >> > > Mike Brooks >> > > CCIE#16027 (R&S) >> > > >> > >> > >> > >> > -- >> > Nick Marus >> > [EMAIL PROTECTED] >> > > > > -- > Nick Marus > [EMAIL PROTECTED] >