s vermill wrote: > > Priscilla Oppenheimer wrote: > > > > s vermill wrote: > > > > > > Larry Letterman wrote: > > > > > > > > In most cases you will only re-write the source mac > address > > > > when traversing > > > > across a L3 device. > > > > > > I don't think that's so. > > > > Did you misplace your comment? > > No. I disagree that a source MAC re-write would be all that > takes place when crossing a L3 device. Host A, sending to an > off-subnet Host B, would use its own MAC as the source and the > L3 device interface MAC as the destination. The L3 device > strips both at ingress. If, in fact, the destination is on a > directly attached shared medium, the source MAC is re-writen to > that of the egress interface. The destination MAC is whatever > the L3 device has in the ARP cache for Host B. Both source and > destination MACs change when crossing a L3 device. Doesn't it > sound like Larry is saying that the source MAC is all that > changes and not the destination MAC? Or maybe I just took that > wrong?
I think that maybe Larry was saying that the only time it would be *necessary* to change the source MAC is when traversing a L3 device. He isn't necessarily saying that only the source MAC would change when crossing one. Sorry Larry. I think that was a mis-read on my part. > > I think his first comment is > > correct, but then a following one is strangely worded. See > below > > > > > A host will have an ARP cache entry > > > for its gateway. That would be the destination MAC. The > > > source MAC would be that of the sending host itself. Using > > its > > > own ARP cache, the gateway would re-write both the source > and > > > destination MAC if the destination was, in fact, directly > > > attached to (or reachable via) another Ethernet interface. > > > If > > > not, and the packet needed to cross some serial WAN link, > both > > > MACs would simply be stripped off. Every L3 device strips > off > > > source and dest. MAC at ingress. Whether or not a new > source > > > and dest. MAC is encapsulated around the IP packet depends > on > > > whether or not the destination is reachable via another > > > Ethernet interface. > > > > Or Token Ring, FDDI, LocalTalk. :-) > > > > > > > > > If you go across a layer 2 network, all > > > > the mac address's > > > > would typically be part of the same broadcast domain and > not > > > > need to be changed. > > > > > > > > If you go across a T1 or Frame it will still be mapped to > or > > > > have an assigned IP Address > > > > that constitutes a layer 3 hop and write its mac address > in > > > > the frame. > > > > Here's where he went astray. As I mentioned earlier, a serial > > interface doesn't have a MAC address and the data-link-layer > > protocols used across serial interfaces don't have MAC > > addresses in them. > > > > The sentence isn't parsable, (sorry Larry!) but may indicate > > some additional misunderstanding. The fact that the next hop > > has a Layer 3 address isn't of major significance when talking > > about forwarding traffic and the addresses that end up in the > > forwarded packet. The IP addresses don't change end-to-end. > MAC > > addresses on LANs change, hop by hop. WANs don't have MAC > > addresses. > > > > Yes, routing protocols exchange next hop info using IP > > addresses. So, if we're considering Ethernet, at some point > the > > source router must have found out the MAC address of the > > destination router using ARP. The router will put its own MAC > > address in the source field and the destination (next hop) > > router's MAC address in the destination field. > > > > In the case of a T1 point-to-point link, a MAC address isn't > > necessary since it's not a shared medium and there's no need > to > > identify which station should receive the frame. There is only > > one other station! > > > > Now, Frame Relay is shared "in the cloud." The DLCI would help > > the L2 switches in the cloud forward the frame correctly. > > Inverse ARP would help the router map a L3 next hop address to > > a DLCI, if I understand it correctly. > > > > Priscilla > > > > > > > > > > > > > > However if I am wrong here, Priscilla or Howard or Chuck > > > > will let me know...:) > > > > > > > > Larry Letterman > > > > Network Engineer > > > > Cisco Systems > > > > > > > > > > > > ----- Original Message ----- > > > > From: "Cisco Newbie" > > > > To: > > > > Sent: Friday, January 31, 2003 11:42 AM > > > > Subject: RE: MAC Address [7:62251] > > > > > > > > > > > > > First, thanks for all that responded. One clarification > > > > that I need address > > > > > is the following: > > > > > > > > > > If I cross a L3 router and the outgoing interface is > > > > something other than > > > > > Ethernet, will the L2 frame show a new MAC address? In > > > > other words, if my > > > > > outgoing interface is say T1 PPP or even a dial-up, > should > > > > I be seeing a new > > > > > MAC address? > > > > > > > > > > Is it only when I cross a L3 device AND my outgoing > > > > interface is a share > > > > > medium like Ethernet that a new MAC address will be > placed > > > > on the frame? > > > > > > > > > > Thanks. > > > > > > > > > > > > > > > > > > > > --------------------------------- > > > > > Do you Yahoo!? > > > > > Yahoo! Mail Plus - Powerful. Affordable. Sign up now > > > > [EMAIL PROTECTED] > > > > > > > > > > > > > > > > > > > > Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=62375&t=62251 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
