From: John Gibson (johngibson1541@yahoo.com)
Date: Thu May 24 2007 - 11:55:27 ART
I was so much into switches that I forgot the
NIC cards of the PCs can ignore 01-00-5E-xxxxxx
frames that do not match the host's ipconfig.
So, the benefits includes saving interrupts to
the host systems' CPUs.
Too much into switches I forgot about hosts.
John
--- Guy Sherr <gsherr@gmail.com> wrote:
> This depends on your application. The 01-00-5E...
> address is slightly
> better than all 1's because it is a multicast
> address -- all stations
> in an L2 network have to pick it up, but only the
> stations that have
> the protocol address enabled on their receivers need
> to refer the
> frame to the PAD. With ffff.ffff.ffff, all stations
> in the broadcast
> domain think the frame belongs to them and the frame
> is referred to
> the PAD anyway. For example, reaching back into
> history, DECNet node
> addresses were encoded 01-00-AA.xx.xx.xx so DECNet
> could ride
> ethernet. The last six octets were created from the
> Phase IV address
> (0-63 followed by 0-1023). The frame would go
> everywhere, but only
> the "right" node would "read" the contents (one side
> benefit: no ARP
> necessary because the address was a computed value).
>
> Refer also to DEC MOP, which only the proper nodes
> could hear for the
> same reason. Bear in mind though that this
> technique for managing
> broadcasts is bordering on ancient, headed for the
> "Arcane and
> Curious" section of networking.
>
> HTH,
> guy
>
> On 5/23/07, johngibson1541@yahoo.com
> <johngibson1541@yahoo.com> wrote:
> > If we don't have IGMP snooping and no CGMP and no
> MAC filtering exists,
> > ethernet multicast address is not more beneficial
> than using
> > ethernet broadcast address ffff.ffff.ffff right?
> Am I thinking straight?
> >
> > So, to live up to the full potential of the
> 01-00-5E-xx-xx-xx design,
> > all switches of all brands should all have some
> IGMP snooping mechanism?
> >
> > Since the multicast addressed frames will be
> flooded out all switch
> > ports by default without CGMP or IGMP snooping ,
> without CGMP or IGMP
> > snooping, 01-00-5E-xx-xx-xx does not cause any
> bandwidth conservation
> > right?
> >
> > With a protected port with multicast blocking,
> ffff.ffff.ffff
> > or 01-00-5E-xx-xx-xx doesn't make a difference
> right?
> >
> >
>
This archive was generated by hypermail 2.1.4 : Fri Jun 01 2007 - 06:55:22 ART