Berkeley CSUA MOTD:Entry 23297
Berkeley CSUA MOTD
 
WIKI | FAQ | Tech FAQ
http://csua.com/feed/
2025/05/24 [General] UID:1000 Activity:popular
5/24    

2001/12/19 [Computer/SW/OS/OsX] UID:23297 Activity:high
12/19   Has anyone ever heard of "MAC multicasting?"  I can only find vague
        references to it;  it's not the "fake" ethernet addresses reserved
        for class D IP addresses.  Apparently Baydel and a few other
        products/applications use it for various kinds of clustering.  I'd
        be grateful for pointers to any documentation.  -John
        \_ you're badly confusing IP addresses and end ethernet MAC addresses.
           \_ No, I'm not.  Nor am I referring to the MAC addresses assigned
              to multicast IPs.  Read the post again.  -John
              \_ John, I don't think there's any "MAC multicasting" that is
                 not multicast IP -  multicast MAC mapping.  I'm not aware of
                 any RFC that describes it.  If a vendor claims it, it's
                 some proprietary shit.  If this is used within a cluster
                 then the boxes can use whatever MAC they want. We do that
                 all the time.  Assign arbitrary multicast MACs to various
                 linecards and let a central route processor send messages
                 to them.  It's standard practice to do that.  But the end user
                 never sees that and has no control over that.  I assume
                 that you're an end user.  -cisco kid
        \_ multicast IP addresses have to be mapped to multicast MAC addresses.
           That's how it works.  E.g., when sending to multicast group
           224.1.1.1 the MAC address should be 0x01005E010101.  Look at
           \_ Yeah looking at software implementations will help you understand.
              But this should already be done by hardware.  I implement this
              kind of stuff in ASICs and microcode.  -cisco kid
           RFC 1112.  -cisco kid
                         \_ hey, you were a friend of mine.
           \- hello. the least sig bit in the first byte [or octet is the
           proper jargon here], of the dst addr is the IEEE802.3 frame is
           a mcast addr. for shorter people: any dst addr that is odd in
           the first byte, is a mcast add. so the bcast is in the mcast
           space and is a mcast addr but wants special handling. so when a
           company gets a range of MACs, there is a bijection to it's space
           of mcast addresses. grep say freeebsd src for etherbroadcastaddr
           and M_BCAST and M_MCAST to see how this is done. ok tnx. --psb
           \_ Yeah looking at software implementations will help you
              understand. But this should already be done by hardware.
              I implement this kind of stuff in ASICs and microcode.
              -cisco kid