Why do systems only map the 23 least significant bits of a class D IP
address to a multicast MAC addresses instead of all 32 bits or at
least 24 bits?

---

Request for Question Clarification by maniac-ga on 11 Jan 2006 18:09 PST

Hello S1m0n,

Are the networks you are using for multicast using Ethernet?

If so, the reason is straight forward (only 23 bits available in the MAC address).

If not, could you describe your situation more clearly so I can search
for an answer more specific to your situation?

Thanks.
  --Maniac

Hello S1m0n,

Never mind the question clarification request. I read the question &
not the title....

As mentioned before, the MAC address on Ethernet has only 23 bits for
multicast address mappings. There are pages such as:
  http://www.hep.ucl.ac.uk/~ytl/multi-cast/addresstranslation_01.html
which has an illustration in the section titled "Ethernet MAC Address
Mapping" which shows that 5 bits are lost from the IP address in the
mapping to the MAC address.

Alternatively
  http://www.cisco.com/univercd/cc/td/doc/cisintwk/ito_doc/ipmulti.htm#xtocid11
a much longer description at Cisco with basically the same illustration.

or
  http://www.tldp.org/HOWTO/Multicast-HOWTO-2.html
an explanation without the nice illustration (scroll to the end for
the mapping explanation).

Search phrases for the answer included:
  IP multicast class D
  IP multicast class D ethernet mapping

Please make a clarification request if some part of the answer is
unclear or is incomplete so I can correct the answer.

  --Maniac

---

Request for Answer Clarification by s1m0n-ga on 12 Jan 2006 20:21 PST

I teach TCP/IP and the question has come up multiple times so I want
an authoritative answer. There are 48 bits in the MAC address but only
32 in the Class D multicst addresses. Since the first byte of the MAC
address uses the least significant bit to indicate multicast, that
removes 8 bits leaving 40 bits to take in the 32 bits of the IP
address. That is not done. Instead, when mapping a Class D multicast
IP address to a MAC address, the four significant bits and the 23
least significnt bits of the IP address. Why not all 32 bits? Since
the four most significant bits are the IP multicast indictor why not
use the remaining 28 bits?

---

Clarification of Answer by maniac-ga on 16 Jan 2006 18:53 PST

Hello S1m0n,

The Internet Assigned Numbers Authority (IANA - http://www.iana.org/)
has reserved half of the 24 bit address range defined for IP
multicast. This reservation (as well as many others) is described at
  http://www.iana.org/assignments/ethernet-numbers
(scroll down to the section starting with "ETHERNET MULTICAST ADDRESSES")
From this reference, each vendor can define a set of Ethernet
multicast addresses - so only the range starting with 01-00-5E can be
used for "Internet Multicast" (and only half of that due to the IANA
reservation). See also RFC 1112 at
  http://www.faqs.org/rfcs/rfc1112.html
which states basically the same restriction (scroll to 6.4).

It is not clear as to WHY there are a set of addresses (using 23
instead of 24 bits) reserved for IANA. The most likely reason is to
reserve a portion of the address space for use in a future protocol.
However, that is an assumption on my part - I cannot find an online
explanation this part of Ethernet history.

There is however, an interesting historical description at
  http://www.intelligraphics.com/articles/ipmulticasting1_article.html
(scroll down to "2.2 Mapping IP-Multicast Addresses to Ethernet MAC Addresses")
where it describes a financial problem with getting additional
Ethernet addresses (the OUI portion) assigned. This is likely why
there is only a single range of addresses (24 / 23 bits) assigned to
IP Multicast on Ethernet (combined with some inertia by vendors /
standards organizations).

I should also note that there are further restrictions on the
multicast addresses you can use on a regular basis. For example, there
are a set of reserved multicast addresses described at
  http://www.iana.org/assignments/multicast-addresses
as well as a list of "unusable" addresses in this reference at Cisco
  http://www.cisco.com/univercd/cc/td/doc/product/webscale/uce/acns52/ldg52/refapp.htm#wp1035126

To recap, there is a single OUI assigned to IP Multicast on Ethernet
(01-00-5E). That reduces the available set of Ethernet addresses from
48 bits to 24 bits. The IANA reserves half of that range, reducing the
available set of addresses to 23 bits. This is supported by standards
established by several standards bodies.

  --Maniac