On Thursday, 02/09/2006 at 11:49 CET, Rob van der Heij <[EMAIL PROTECTED]>
wrote:
> On 2/9/06, Alan Altmark <[EMAIL PROTECTED]> wrote:
>
> > It's a design philosophy.  Displaying the access list for a VSWITCH is
> > difficult if it is a combination of GRANTs plus NICDEFs.  We don't
want to
> > search the directory every time someone does a QUERY.  And displaying
an
>
> I figured the design was broken rather than the code, otherwise I
> would have tried a PMR against it. I fear Chuckie has been smoking the
> same stuff as Miguel...
>
> The need for a QUERY is not obvious to me. What's the benefit of being
> able to tell what other servers could connect to the same network as
> you did ? The only purpose now is to check the superfluous GRANT.
> Network design needs some careful planning anyway and now you have the
> same people define the same aspect in different places.

As I say, a different philosophy.  We chose to have an explicit access
list for guest LANs and VSWITCHes rather than an implicit access list
based on virtual machine configurations.  And we didn't want to treat a
dynamic COUPLE any differently than a NICDEF.

And I happen to think that the directory should be restricted to defining
virtual machine configurations, with only minimal involvement in access
rights.  I'd like to be able to implement a remote directory access
service and still be able to locally administer access rights.

> Yes, the way I see it when you drop your NIC you can COUPLE that one
> again to the LAN defined in the directory - only that one. I warned
> against the risk of DoS with Guest LAN and that was not fixed with the
> VSWITCH.

Sorry, Rob, but what does DoS have to do with access rights?

> Sure, access control through the CP directory (and passwords) is
> different from using an ESM and it should be.

I respectfully disagree, Sir Rob.  I believe that the access control model
should be the same, whether CP or an ESM is in control.  That is: "Here is
a restricted system widget and the list of user who can access it."

> With an ESM you can
> separate functions and manage your access control much better. When
> the CP directory is the only control, then it is unwise to force the
> adminstrator to define the same thing in two different places because
> that causes stale authorisations.

Agreed.

The industry tells us that security is Very Important.  Not just the
security itself, but the management thereof.  We do make an effort to
maintain the balance between convenience and security, but it ain't easy.
("VLAN ANY", anyone?)

Alan Altmark
z/VM Development
IBM Endicott

----------------------------------------------------------------------
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