On Dec 13, 2012, at 7:46 AM, Melinda Shore wrote:

> Likely of interest to the OPS community.
> 
> -------- Original Message --------
> Subject:      Feedback Requested: Proposed IEEE RAC OUI Tier Restructuring
> Date:         Wed, 12 Dec 2012 19:37:23 -0800
> From:         IAB Chair <[email protected]>
> To:   [email protected] <[email protected]>
> 
> The IEEE Registration Authority Committee (RAC) has requested IETF
> feedback on a proposal for restructuring of the Organizational Unique
> Identifier (OUI) within the IEEE 802 Medium Access Control (MAC)
> Address.  Information on the proposal is available here:
> 
> http://www.ietf.org/iesg/ieee/20120725/RAC_Virtualization_July2012.pdf
> 
> http://ieee802.org/secmail/msg15524.html
> 
> Comments should be sent to Glenn Parsons <[email protected]
> <mailto:[email protected]>> and cc’d to [email protected] <mailto:[email protected]>.
> 
> _______________________________________________
> OPSAWG mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/opsawg


My first thought on scanning the deck is that the problem is fundamentally one 
of scaling. The reason that we structure addressing and routing as we do in IP 
is to make addressing and routing scale. The Ethernet MAC was designed to 
identify a NIC on a physical wire, and by extension in a VLAN. It seems to me 
that the VM problem they are having derives primarily from the confusion of a 
network attachment point on a broadcast LAN with a large scale routing 
architecture.

As noted, one approach is a DHCP-like assignment of MAC addresses. This doesn't 
do away with physical addresses; the DHCP-like request would come from the 
physical address, and the response to it, assigning a virtual address to a VM. 

But it frankly seems a lot simpler to use RFC 4941 addresses layered onto the 
physical addresses and as a result a scalable routing infrastructure, rather 
than trying to do large scale host routing.

My $0.02. 
_______________________________________________
OPSAWG mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/opsawg

Reply via email to