|
Posted by Michelot on November 30, 2007, 4:25 am
If you were Registered and logged in, you could reply and use other advanced thread options
Bonjour,
I am thinking a curious thing I had never considered before, or simply
forgot.
(1) A basic hub has no MAC layer, so it can look at Ethernet
addresses.
(2) A basic switch is a layer 2 interconnection device, so it can
consider Ethernet addresses.
(3) A basic router is a layer 3 interconnection device. It has a MAC
layer like a switch but, besides, each port has it own Ethernet
address.
What would be the difference between a "switch MAC layer" and a
"router MAC layer" that makes in one case we don't have an Ehernet
address and in the other case we have an Ethernet address? Perhaps the
MAC layer can be divided in several sublayers.
Thanks for your advice,
Michelot
|
 
| |
Posted by Michelot on November 30, 2007, 9:46 am
If you were Registered and logged in, you could reply and use other advanced thread options
Bonjour Michelot,
> What would be the difference between a "switch MAC layer" and a
> "router MAC layer" that makes in one case we don't have an Ehernet
> address and in the other case we have an Ethernet address? Perhaps the
> MAC layer can be divided in several sublayers.
The word "bridge" has really its importance for that. Routers and
switches use both the MAC layer but not for the same services.
- Routers use MAC layer in the "vertical" sense (from MAC to level 3,
and level 3 to MAC),
- Switches use MAC layer in a "transverse" sense (from MAC to MAC as a
relay).
First services (for basic routers) are those specified by the OSI
model, and second services (for basic switches) are the internal
sublayer services specified by 802.1D.
Do you agree that,
best regards,
Michelot
|
|
Posted by Michelot on November 30, 2007, 9:50 am
If you were Registered and logged in, you could reply and use other advanced thread options
> What would be the difference between a "switch MAC layer" and a
> "router MAC layer" that makes in one case we don't have an Ehernet
> address and in the other case we have an Ethernet address? Perhaps the
> MAC layer can be divided in several sublayers.
The word "bridge" has really its importance for that. Routers and
switches use both the MAC layer but not for the same services.
- Routers use MAC layer in the "vertical" sense (from MAC to level 3,
and level 3 to MAC),
- Switches use MAC layer in a "transverse" sense (from MAC to MAC as a
relay).
First services (for basic routers) are those specified by the OSI
model, and second services (for basic switches) are the internal
sublayer services specified by 802.1D.
Do you agree that?
best regards,
Michelot
|
|
Posted by Albert Manfredi on November 30, 2007, 8:00 pm
If you were Registered and logged in, you could reply and use other advanced thread options > Bonjour Michelot,
>
> > What would be the difference between a "switch MAC layer" and a
> > "router MAC layer" that makes in one case we don't have an Ehernet
> > address and in the other case we have an Ethernet address? Perhaps the
> > MAC layer can be divided in several sublayers.
>
> The word "bridge" has really its importance for that. Routers and
> switches use both the MAC layer but not for the same services.
>
> - Routers use MAC layer in the "vertical" sense (from MAC to level 3,
> and level 3 to MAC),
> - Switches use MAC layer in a "transverse" sense (from MAC to MAC as a
> relay).
>
> First services (for basic routers) are those specified by the OSI
> model, and second services (for basic switches) are the internal
> sublayer services specified by 802.1D.
>
> Do you agree that?
Bonsoir Michelot,
Hubs, i.e. repeaters, should not consider MAC addresses at all. An
incoming frame is flooded to all other interfaces of the hub,
emulating the experience of parallel-connected hosts in a coax
Ethernet.
The most basic bridges also operate this way, even if they have a MAC
address for every port. The original purpose of bridges was to tie
together different Ethernet media into a single L2 net, or just extend
an Ethernet beyond the limited extensions allowable with simple
repeaters. So the MAC addresses of these bridges didn't play a role in
basic frame forwarding at all. (Never shows up in frames sent between
hosts, for example.)
The special category of Layer 2 switches called "learning bridges,"
which describes most L2 switches these days, do use the MAC DA of an
incoming frame, to determine to which port that frame should be
"routed." I think that's what you are describing as "transverse." This
is a special case.
Routers use the MAC layer just as any host does. The MAC address is
only used to find that router's interface in the L2 net, that's all.
What the router does afterwards, to forward the frame, is independent
of MAC address considerations.
In short, I think it's safe to say that NORMALLY, a MAC address is
only used to identify a host's interface in the L2 network. As such,
repeaters and bridges, in their most basic forms, don't use MAC
addresses at all in forwarding frames.
If you need to manage a bridge, then yes, the MAC address is used, but
that's only because the managed bridge is behaving like a host in this
case. And learning bridges use MAC addresses for their special trick.
And link aggregation uses MAC addresses to bundle together a group of
switch interfaces. But I think it helps to consider that these are all
special cases.
Albert
|
|
Posted by Michelot on December 1, 2007, 5:48 pm
If you were Registered and logged in, you could reply and use other advanced thread options
Thanks for your development and extension of my questionning. At 2.00
am, don't you sleep in your country?
> The special category of Layer 2 switches called "learning bridges,"
> which describes most L2 switches these days, do use the MAC DA of an
> incoming frame, to determine to which port that frame should be
> "routed." I think that's what you are describing as "transverse." This
> is a special case.
It was my philosophic investigation, becoming aware of the different
use of the MAC layer.
As you said, for the strict forwarding function, the switch interface
doesn't need to be identified by a MAC address, although it uses the
MAC layer.
This is described in 802.1D, in the figure 7-3.
In the MAC layer, we have (1) the MAC Entity and (2) the MAC Relay
Entity.
(1) For each bridge port, the MAC Entity handles the MAC protocol and
exchange "MAC Services" with the higher layer (for management, STP,
LACP...). The "MAC services" use a MAC adress that identies the port.
(2) The MAC Relay Entity handles functions of relaying frames between
ports, filtering, learning. It uses "Internal Sublayer Services"
provided by the separate MAC Entities of each port. It is what I
called the "transverse" sense. And this sense doesn't refer to a
specific MAC address of a port.
So there are effectively 2 kinds to consider the MAC layer. And,
certainly, we can find for a router an IP Relay Entity, with "IP
internal Sublayer Services".
Best regards,
Michelot
|
| Similar Threads | Posted | | Low-end Layer 3 switch? | February 14, 2006, 6:19 pm |
| Layer 2 switch vs Layer 3 switch | October 5, 2007, 1:28 pm |
| HELP!A SURVEY ABOUT THE LAYER-3 SWITCH | February 17, 2006, 12:34 am |
| Layer 3 switch model? | March 30, 2007, 1:20 pm |
| Routing between VLANs using a Layer 3 Switch | March 24, 2005, 8:39 am |
| Placement of Layer 3 3COM switch query | May 30, 2005, 9:56 pm |
| FS/FA: $160.00 Extreme Networks Summit48 Switch w/basic layer 3 (works great) 2944 | August 9, 2005, 10:31 am |
| about of router and switch | July 12, 2007, 3:18 am |
| IP layer and layer 2 IGMP Snooping | June 7, 2008, 1:28 am |
| Gigabit core router/switch | July 26, 2005, 7:15 am |
|
|
Yahoo! 
Google 
Windows Live 
del.icio.us 
digg 
Netscape 
XML site map