Brian,
The "balanced hash" approach does not preclude you from having a set
of symbolic links using the first two letters of the account name:
AFS account "dexter" id is 1026, 1026%64=2, hence /afs/@cell/u/u2/
fs mkm /afs/@cell/u/u2/dexter user.dexter
ln -s /afs/@cell/u/u2/dexter /afs/@cell/home/dexter
ln -s /afs/@cell/u/u2/dexter /afs/@cell/user/d/e/dexter
This addresses both issues of balancing the user volumes across
the "mid-level" directories (eg u0 through u63) and having exactly
the same "first two letters path" of the UoM method.
--
paul http://acm.org/~mpb
"Ooh! What does this button do?" --Deedee
http://www.planet.net/vanman/dexter/whatdou.wav
>On Thu, 16 Apr 1998, Rich Sudlow wrote:
>
>| The University of Michigan directory structure (which is the same one being
>| proposed) is indeed the right one to use. Given the afs_id you know
>| exactly where the users home directory. This is exactly what you want,
>| no lookups.
To which Brian Spolarich responded:
bws> Its also a lot easier for users to understand. I'd recommend against
bws>the 'balanced hash' approach. It may be technically correct, but probably
bws>less practical. It would seem to me that its unlikely that you're going
bws>to encounter such a glut of userids with the same first two characters
bws>that you'd encounter directory lookup problems. If you do find such a
bws>concentration emerging for some reason, you can always disallow any more
bws>usernames in that bucket. :-)
>
> -bws
>
>--
>Brian W. Spolarich - ANS Communications - [EMAIL PROTECTED] - 734-214-7311
> "Not a whit, we defy augury." - Hamlet, V, ii
