Greg Freemyer wrote:
On Fri, Jan 6, 2012 at 1:48 PM, James Knott<james.knott@rogers.com> wrote:
Greg Freemyer wrote:
If you have a full class A (ie. /8 or 10.x.x.x) then you conceivably have 10's of thousands (or more) MACs to potentially keep up with and setup spanning trees for.
Wait 'till you start working with IPv6 networks, where you can have up to 2^64 devices on a subnet!
Of course, practical considerations may limit that somewhat. ;-)
I have a /56 IPv6 subnet, which can be split into 256 /64 subnets. I'm a IPv6 neophyte, but I have wondered about that.
Since IPv6 is 128 bits I think, a single /64 subnet is as big as the public facing Internet is today. (ie. Ignoring private IP space like 10.x.x.x).
Actually, it's the entire IPv4 address space squared! IPv4 has 2^32 or about 4 billion addresses. 2^64 is 18.4 quintillion. My /56 subnet is 2^72 or 4.7 sextillion, but I haven't used them all yet. ;-)
I have a hard time fathoming how something that big can be managed without some form of hierarchy.
Actually, the IPv6 address space is designed to be a hierarchy, so as to reduce routing table size. Also, most subnets will actually have only a tiny fraction of the possible addresses. The current practice, with recent versions of Linux, Windows etc., is to have at least 3 addresses per device. Those are the link local address, starting with FE80, which all IPv6 devices must have. Then there's a MAC based address on the subnet and due to privacy concerns, one on the subnet using a 64 bit random number, in place of the MAC based address. The MAC address, when used as part of an IPv6 address has FFFE stuck in the middle to fill it out to 64 bits. The reason for that random number address is that the MAC based one is tied to a specific computer and some people think that may be a security issue. So the random number address is used for outgoing connections and the MAC based, incoming. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org