SLAAC can operate under two models. 1) It will generate based on the hardware MAC address. 2) It will generate based on "some other token".
Microsoft uses "Some other token". So, there shouldn't be a conflict with MAC addresses under that platform. -----Original Message----- From: Ken Schaefer [mailto:[email protected]] Sent: Tuesday, July 13, 2010 12:06 AM To: NT System Admin Issues Subject: RE: DHCPv6 So SLAAC will only work if you have unique MAC addresses? If you use Hyper-V, then the pool of MAC addresses assigned to the guests is based off a pool generated from the host's IP address. If you build servers in a build factory, then you'll end up with duplicate MAC addresses for your guests. Cheers Ken -----Original Message----- From: Ben Scott [mailto:[email protected]] Sent: Tuesday, 13 July 2010 11:00 AM To: NT System Admin Issues Subject: Re: DHCPv6 On Mon, Jul 12, 2010 at 10:29 PM, Kurt Buff <[email protected]> wrote: > No familiarity with DHCPv6, so an ignorant question... This is currently the subject of holy wars on forums such as NANOG. An IPv6 node can discover the network number, network mask, and local routers by using router solicitation. This is part of the core IP protocol, and in theory should be part of every implementation. The IPv6 node can then use its MAC address to generate a unique address on the local network (this is called SLAAC (StateLess Address Auto-Configuration)). So an IPv6 node can get a working network layer on any network, without DHCPv6. However, you still need DHCPv6 to find out things like DNS servers. So SLAAC is only good for layer 3, not for higher layer stuff. This has lead to a feud between those who think IPv6 address assignment should work just like IPv4 -- via DHCP -- since that's what everyone's infrastructure is built around, and thus SLAAC is just a waste of resources, vs those who think addresses should come from SLAAC and DHCPv6 should only be used to discover higher layer stuff. Implementations behave according to which armed camp they align with. Things haven't shaken out yet. Until they do, I expect IPv6 client-vs-network interoperability (i.e., "How do I configure my pee sea for your net work?") to be a clusterfsck. ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ <http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/> ~ ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ <http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/> ~
