On Mon, 25 Sep 2000 19:11:25 -0500 (Central Daylight Time), Glenn Cole
<[log in to unmask]> wrote:

>c|net's NEWS.COM has more on SuperDome, at
>
>   http://news.cnet.com/news/0-1003-200-2757613.html
>
>One interesting item:
>
>   HP is changing payment plans so that customers' payments go up or down
>   depending on how much of the machine's computing capacity is being used.
>   Previously, HP only offered the option to buy a server with unused
>   processors, paying only when a customer needed to fire them up.
>
>Still, the machine sounds too darned complex.  As if sysadmin of HP-UX
>was not complex enough already.  (Jeff?  Richard?)
>

I'm not Jeff or Richard, but I'm expecting that Superdome admin itself
isn't going to be all that much more complex. It's usually HP's
"layered products", like Mirror/UX and MC/Service Guard, which can
make things rough. Otherwise, it's just a big machine with lots of I/O
slots and memory carriers. This is usually pretty transparent to the
admin in general.

The biggest potential area for new complexity (and hence some probably
grpwing pains among admins) is going to be the virtual partitioning
scheme, whereby you can turn your single physical machine with
multiple CPUs into a number of virtual machines, each running their
own copy of the OS (currently, only HP-UX is qualified for this. HP
promises to include Linux and most likely Windows 2000 or it's
descendants in the future). IBM has been doing this on their big big
iron for decades now, so it's not a breakthrough science or anything
.. the theory is well-known with a lot of field experience. But it
wwill take a bit for HP developers, HP support, and HP admins to get
the hang of it.