> > Except there's nothing supplied that's capable of exploiting it, so
it's
> > pretty useless. Attaching a virtualized VT220 to the HMC isn't all
that
> > useful either given the arcane and fragile nature of the HMC.
> 1. z/VM 5.3 allows you to ATTACH the real HMC ASCII console to a
guest. 

I guess I don't consider the HMC a component that common citizens should
be permitted to manipulate. It certainly isn't something that the
unwashed masses should be permitted to touch, and it's not part of the
VM environment. 

> I don't understand your point about nothing exploiting it;
> it's for people.  

There are no tools within a VM system that can drive or manipulate data
streams to the virtualized VT220 console. The HMC is not a VM component
(in fact, it is deliberately OUTSIDE the scope of any operating system
by design), and the assumption that anyone needing that VT220 console
access should ALSO have either a) physical access to the HMC or b)
network access to the HMC is optimistic, IMHO. 

It's not unusual these days for the machine to be several hundred miles
away from the people, and/or physically inaccessible to even the systems
people. ATTACHing a process that runs only on a closed box with no
external integration points isn't (at least IMHO) exploitable. I'd also
really hesitate to allow your average Unix administrator access to
something like the HMC; the access controls exposed in the HMC UI are
just not manageable at any scale. If there's a way to integrate access
control via external sources into the HMC *that IBM is willing to
support*, I'd sure like to know where to read about it.

>From a usability standpoint, it'd be my expectation to AT MINIMUM be
able to do SSL-protected telnet to a concentrator of some time and
select a connected system to interact with. 

> 2. I don't understand the "arcane and fragile" comment.  [snip]
> Maybe you're referring to the "F-R-A-G-I-L-E" on the shipping
container?
> That's the country of origin, pronounced "fra-gee-lay".  :-)

Nyet. 

It is trivially easy to close a window and lose it in the HMC GUI. It is
trivially easy to generate a situation where you cannot open an
important HMC application you need to correct a problem because the HMC
GUI insists that a window for that application is already open
somewhere. It is trivially easy to cause assorted remote access failures
to the HMC GUI from external sources. It is trivially easy to replay
session streams back to an HMC if it is network accessible. I could
continue, but confusing UIs or those kinds of reliability issues
constitute fragility. 

I don't disagree that the HMC has improved a LOT since the MP2k days. I
do contest that it constitutes a good design management goal to have it
in the loop for things that are in normal operations for virtual
machines. An HMC can do way too much harm to have that be attractive. 

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to