-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Sat, Aug 23, 2008 at 10:07:58AM -0700, Eric P. Melbardis wrote:
> Hi
>
> Testing on my home system, looking good... will try on my office system Monday.
>
> Note:
> I am assuming that the uuid will mostly be used as a primary
> key on records that will exist in replicated databases to guarantee
> uniqueness. The problem that is see with the current uuid functions
> for this purpose is that the generated uuid is very random causing
> great performance issues with the index [...]
> Hence the older style of uuid, based on mac
> addresses and time tend to generate more sequential uuid's,
> at least when inserting a bunch of records at once.
>
> In other words it would be nice to have the old fashioned style of uuid's.
Note that in the days of virtual machines, those more or less "invent"
their MACs. Since they just have to be unique in the subnet (or
"broadcast domain"), collisions on a more global scale will go
unnoticed. But they might bite you in a replicated database scenario.
I wouldn't trust "old fashioned uuids" anymore.
Regards
- -- tomás
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFIsRk8Bcgs9XrR2kYRAkX6AJ9EoGka6NZ8dGijynYr/ujogStqfgCdEhU6
b1yyo+YNpmeqJ2IRMqowMHE=
=iZHA
-----END PGP SIGNATURE-----
