Re: Commands for Limited Users

Well, "back in the day", it was a big deal to place a command line as
part of our menuing system, and in many cases as an option in many
programs. So, limit capabilities is quite important in our situation. 

I am sure there are many ways to track many different items to be able
to control things on a command by command basis, but then I would be
repeating another level of what we have been forced to do by the
government, and then the auditors already. However, when things start
slowing down, I'm sure we'll look at additional minutiae...

Dave 

-----Original Message-----
From: security400-bounces@xxxxxxxxxxxx
[mailto:security400-bounces@xxxxxxxxxxxx] On Behalf Of Phil Ashe
Sent: Wednesday, September 06, 2006 4:22 PM
To: Security Administration on the AS400 / iSeries
Subject: Re: [Security400] Commands for Limited Users

I wanted to clarify some things about limit capabilities and command
security.

Limit capabilities is part of the legacy support for systems at security
level 10 or 20. It is a tool to solve a problem you probably don't have.
It is designed to enforce a menu security system on terminals when you
have no other security.

At security level 20 (security level 10 hasn't been supported for
years), you aren't interested in managing object-level security and you
have a very small set of tools to control user actions. In this
environment, every user has *ALLOBJ rights and has the ability to run
every command. There is no "security" except that which is provided by
limit capabilities and menus. Limit capabilities and menus provide the
means of controlling what users can do in an ad hoc fashion. 

At security level 20, you don't have object-level security. As John Earl
pointed out, limit capabilities won't secure commands from non-command
line interfaces, including commands from remote servers and within
programs. It wasn't meant to. Limit capabilities was designed for menu
security and not object-level security. This behavior outside of menus
is to be expected. 

Neither FTP nor Rexec is a classic OS/400 command line interface managed
by limit capabilities functionality. Many of these types of interfaces
didn't exist when the limit capabilities support was created. If you
want to control access to commands in all interfaces (programs
included), you need an object-level security scheme.

At security level 30 or higher, you have other options for controlling
the actions of users, including object-level security. You can simulate
the functionality of limit capabilities by a combination of command
proxies, programs, specialized signon screens, and object-level
security. 

The benefit for managing limited capabilities is very small at security
level 30 or higher, especially since you can provide for similar
functionality through other means. For a similar amount of effort, you
can turn on action auditing for users and determine actual commands
being run. After an analysis of that data, you then hone your object
security plan. Well-managed object security provides a much bigger
benefit than limited capabilities. 

CHGPWD shows some of the problems with limited capabilities. The CHGPWD
command has no parameters. It's a command that almost everybody should
have access to, but it is shipped with "Allow limited user" set to
"*NO". 

Why is it OK for a capabilities-limited user to access CHGPWD by typing
"8" on the User Tasks menu (GO USER) but not type out "CHGPWD" on a
command line on the same menu? It makes no sense.
 
Limit capabilities doesn't provide real security. It doesn't provide
consistent security. Any auditor who insists on using limit capabilities
doesn't understand the issues. I could argue that its use does not
constitute a "Best Practice" in iSeries security.

Phil Ashe
NetIQ (A division of Attachmate)
1233 West Loop South, Suite 1800 | Houston, TX 77027 USA
713.418.5279 phone
phil.ashe@xxxxxxxxxxxxxx
www.netiq.com