Subject Re: [Firebird-Architect] Digest Number 964
Author Jim Starkey
David Johnson wrote:

>On Wed, 2005-03-30 at 06:53 +0000,
>>This is a point I didn't understand how do you get an unique ID based
>>MAC or IP Address for unconnected DB's that is guaranteed unique,
>>made with human intervention ?
>MAC (Media Access Controller) address is guaranteed unique for any
>network media device. It is encoded in the EPROM or an electronic
>serial number chip on every network controller device built.

Many media devices allow the MAC addresses to be set. DECnet, in
particular, requires such devices. Virtually all NAT routers allow
override of MAC addresses.

In theory, MAC address are unique. Practice, however, trumps theory.

>It is a requirement for the functioning of the ethernet CS/CD system,
>and any other networking system for that matter. If you MAC address is
>not unique, then you will get crosstalk at the network hardware level.
>This is a really bad thing - it's even worse than two machines on the
>same network having the same IP address.
Yup and nope, respectively. Doesn't work, but you can do it.

>Systems without a MAC often use the serial number of the BIOS or CPU as
>the starting point. They are not guaranteed to be as unique, but they
>are the source of the "sufficiently unique" comment that non-MAC based
>methods of computing UUID's are constrained by.
Who care's about a machine not on a network?

>To be fair, Jim's mechanism looks to me like a variant of the MAC based
>UUID, without the randomization code that adds a level of safety to the
>uniqueness of the UUID for global scale applications.
No, not at all. Not even close. Not even the same ecological niche.