[Beowulf] scheduler policy design

[Mark Hahn]

>> hell no.  only a few users even have a guess about how long their
>> job will run, let alone what it will actually do (in mem or IO usage).
>
> Our users are roughly split into 3 groups, those who have a reasonable idea of
> how long their job will run,

we have something like these groupings as well, but if you're defining 
their mean clue level, I'm claiming they're heteroscedastic enough
to make predictions basically useless...

> those who haven't enough experience to guess (or
> to know that they can't) and those who run jobs whose runtime is extremely
> dependant on the data and can't be predicted beforehand.

of our 1800 users, I'd say around 1% can predict job behavior to within 15%.
everyone else might be able to guess within an order of magnitude (say,
base 20 ;)

> But then some of our users haven't had to use the command line before.

I don't know for sure, but I think our commandline-savvy-ness is approx:
 	30% old hands at *nix, edit fortran with ed, etc.
 	40% functional but not completely comfortable.
 	30% call directories 'folders' and wish they could click more.